Slashdot Mirror
How to Test Your T1?
lawpoop asks: "We have a T1 line for our building with a local ISP. Right now, we're looking for competitive bids from different companies. The local guy is offering a good price, but the larger guys are saying he may be overselling the T1 service through a DS-3. He swears he's not. So, how do I tell? The sales guys say 'There's bandwidth meters on the web,' but they fail to mention exactly how I can tell if I have a true T1. I've tried a half-dozen bandwith meters on various websites, and the results are highly variable. We've gotten 300-900 Kbps. Each site has disclaimers as to internet traffic, time of day, etc. Furthermore, we split the T1 out over a hub with two other tenants in the building. I'm coming through from behind that hub. How can I tell for certain that I'm getting a full T1? A service tech with a line tester? Any dead-on bandwith meters? What would an oversold T1 read out to be as compared to a true T1? If the larger guys are trying to scare me to their service with stories of oversold T1s, I need to know that they aren't doing it also!"
MRTG
Maybe you could use a packet sniffer and see if you are getting other stuff on your line. If you are, then it's not a true T1.
24 channels of 64 kbits apiece. We sell T1s to customers, and if one of them wanted a util to test their bandwidth (the full 1.544 mbit) they could download a file from an ftp right at our pop. Or, have them ping flood you... use something like mrtg to graph the results, etc.
FLR
1. Start stopwatch
2. Find XXX password site
3. Use aforementioned password
4. D/L 100Gigs of pr0n
5. Stop stopwatch
6. Do the math
DSLreports.com is the only site you'll EVER need for that purpose. Just go to the tools section. It will tell you your up and down, how it compares to other types of connections, and you can enter what city you live in and your ISP so that it can compare your speeds between other people all over the country with every type of connection. It's really an awesome, quick little tool, go check it out.
Finally, math books without any of that base 6 crap in them.
If ISP is a business, it will try to squeeze most out of T1. If you host stuff it might be good.
Otherwise getting business ADSL/Cable/Wireless would be better idea. Having your own T1 linked to Backbone is reallly nice idea, if the sharing parties don't abuse it. Its only around 150k downstream , think of it that way.
Its best to put switch that would have traffic counter on the ports so if someone is abusing
the T1, everyone else would be able to see that.
Setup a mail script to query switch/router, and send daily mails to respective parties.
p.
I've noticed with a few of our local smaller broadband companies that they will have a diagnostic bandwidth meter on their website.
:)
The advantage to that is there is no 'internet traffic' delay to speak of because its basically a direct connection from one end of the line to the other. They've found it to be an invaluable diagnostic tool for tech support.
I'm guessing not to many larger companies are going to do this, that and "fixing" the meter to their advantage is always a possibility
First of all, you probably have a t1, I have no doubt about that. A t1 is a t1 is a t1, there is no questions about that. If you have a t1 then I am sure that you could transfer 1.544mbps over that line to the other side. Now, after you get past the point directly on the other side that is an entirely different question, AND ONE THAT YOU DIDN'T ask. Unless you have that t1 connected to somewhere with so much aggregate bandwidth that you're head will spin I seriously doubt that you will ever get every single kbps that a T1 is good for. If you can sustain 200kb/s downloads or get close, then you have just about all you're going to get.
Any decent router or firewall (which you will likely need to purchase or lease anyway.) Will have this capability built in.
m rtg/mr tg.html
Then there's always MRTG
http://people.ee.ethz.ch/~oetiker/webtools/
fslg503-985-8686503-985-8686503-985-8686503-985-8
download WonderShaper from freshmeat.net. It has the built-in capability to test T1s and T3s for bandwith, throttling, retries, drops, and runt packet stats.
Are you asking how to make sure you're getting a full T1 worth of transport to your ISP, or are you asking how to tell if the ISP is overselling their upstream bandwidth? If the latter, the answer is yes. The problem is if the upstream load of your ISP peaks out. Any ISP you go with is "oversubscribed". The problem is if your ISP is peaking to their upstream. A more appropriate comparison to make is the level of diversity and redundancy of the upstream.
If you're getting T1 service, it should be possible to borrow a CSU/DSU, put it in remote-loopback, and make sure you have a full T1 of bandwidth. Or, if you own the router, you can just look at the statistics...
If you're talking about getting ISP service with "T1" equivalent bandwidth, that's a different story. You wouldn't be able to tell if the guy has "oversubscribed" you unless you find other buyers of the sevice and generate enough traffic to load down the DS-3.
To prevent getting burned, make sure your SLA clearly states the bandwidth you are expecting, and the means by which that is measured.
Cytlid has a good point - you get a T-1 from the phone company (or a reseller/CLEC) and it either IS a T-1 or it IS NOT.
I suspect that you're asking how you can tell whether or not your ISP is selling 50 million T-1 lines when he himself only has a T-3 connection with the rest of the world.
I think the simplest way is to ask. Talk to the sales engineers who work for the larger guys - tell them "Ok, you're trying to scare me away from a smaller vendor...how can I prove for myself how he's configured?" Ask the small guy "This looks like a really good deal...can you demonstrate to me I'll get X level of performance?"
Well, usually, you're buying a T1 from your location to the ISP or hosting company or whoever. 99 times out of 100, you're going to have a full T1 from your place to theirs (ie: 24 channels of 64kbps, or 1.544kbps). The 1% is most likely going to be some unscrupulous ISP; I've heard stories, but never seen a T1 sold to a customer that wasn't a full on "T1".
Now, as for your bandwidth, that's a different story. It is accepted practice to have oversubscription on your network; ISPs simply don't have the money to provide a full, balls-to-the-wall, 1.544mbps connection to the net for every single one of their customers. Local loop charges for simply a T1 from their office to yours starts at around $200 (in Alabama), and that's only if you use a CLEC. The bandwidth is what you're wondering about, and quite frankly, without having someone in the know inside your ISP, you will *never* find out how oversubscribed they are. What you pay above and beyond local loop charges are bandwidth access charges; you're actually paying for internet access at that point. Roughly analogous to paying for a phone line (local loop), then paying for dialup internet service (the T1, in this analogy).
Basically, unless you're buying a DS3 (44.762mbps) or above, you will never ever get committed, 100% full bandwidth on demand all the time.
Van Jacobson's pathchar can measure this. http://www.caida.org/tools/utilities/others/pathch ar/
Another variant that measures available bandwidth was just presented at SIGCOMM 2002...
http://www.cis.udel.edu/~dovrolis/bwmeter.html
Couples periodic packet trains with queuing delay analysis. Does a binary search over a range of periods to zero in on the bandwidth. Run the tool continually and look for the maximum.
I think you found the way, but you forgot to include the address for your site.
Slackware: old school feel, new school gear.
Its just a matter of how much it is over sold. If the "Big Guys" are try to tell you they dont over sell they are lying. Everyone does, playing the averages are how this bussiness works. Ask to see some MRTG graphs of thier gateways b/w. See if thier heads are bonking against the top very often. BTW chances are that the small isp will treat you very well while the super available mega corp sales man soon be replaced with a touch tone menu. Ask for refrences and call them.
Easy. Click here.
Thank god for Microsoft.
A line tester will tell you if you've got a physical T1 but can't tell you how much upstream bandwidth is available. Unless the provider is willing to give you SNMP read access to their network (uh, yeah, right, that'll happen), MRTG isn't going to be very helpful either.
The best and most reliable tool for hop-by-hop bandwidth determination we've found is PCHAR:
http://www.employees.org/~bmah/Software/pchar/
Now!
I learned all I know from other members of my LUG. What I didn't know I could ask.
There is no way for you to determine whether the small guy is overselling his uplink without getting into his data center and doing an audit of his equipment -- something he'd have to be crazy to allow (*I* don't want a T1 from someone who lets potential customers do that!).
However, as other people have noted, after installation it's fairly easy to measure the bandwidth you're actually seeing. Telltale signs of oversold uplinks are things like vastly better performance at 3am than 3pm.
All in all, it's not necessarily a bad thing to have a (slightly) oversold uplink, as long as it is constantly monitored and upgraded if/when end users' aggregate usage is more than 75% for any length of time.
Cheers
-b
As it was stated 24 channels at 64k a peice for a T1. Now if this is point to point, I am not aware of how he is "overselling" his bandwidth. You simply cannot do that in a point to point as the cross connects from CO to CO are directly routed. Now if it was frame-relay, then thats a different story because then you have to worry about how much CIR you purchased. CIR is commited information rate, basically how much traffic is guranteed that will not be marked discard eligable. Once a packet is marked DE, it potentially can be dropped due to congestion in the frame switch thus it goes deep with FECN and BECN, etc.
Overall to me it sounds like he is feeding you a line.
But if you REALLY wanted to test, I suppose you could look into a TBird but then you would need to contacts in the local terminal to run head to head with you.
The real test, frankly, is to get bandwidth from someone with heavy-duty backbone connection (e.g. AT&T) and just plain hammer it with mondo file transfers scattered across the day. If your transfer times are varying with Net traffic periods, your ISP is the bottleneck.
You might be able to get similar information cheaper by doing repeated traceroutes and logging the delay between the ISP and the next router up, which would indicate the queueing delay at the ISP's routers.
Lacking <sarcasm> tags,
An ISP that I know fairly well (*cough* work for *cough*) oversells bandwidth. They use mrtg (as has been suggested elsewhere) and any time a network segment reaches 80% utilization at any point in the day, three days in a row, that segment is upgraded.
Seems to work quite well to me, but maybe I'm biased. Try an get a conference with the techs (see if you can talk to their network monitoring team) and see if they employ a similar practice.
MRTG doesn't test jack $hit. You have to measure between your serial router and the upstream network at the ISPs side. See if they will install netiq qcheck, and measure udp streaming.
http://www.netiq.com/qcheck/howqcheckworks.asp
I don't think you really understand the difference between a T1 and allocated bandwidth. Just be happy it is up and that you are getting good latency. A few tools you may use are: traceroute, ping, and ftp.
I would have thought this type of question would have been asked about 5 years ago. Guess things don't change much.
It's an old ISP myth. Everybody "oversells" their connections at some point in the stream. In the early days, this sort of thing was an issue, a small ISP would but a certain upstream bandwidth (usually one or two connections) and then sell pieces of it until they had sold more downstream than they had upstream. In practice, it worked well, since few people ever use their maximum bandwidth constantly. A few were fastidious about buying upstream bandwidth in exact proportion to what they sold downstream. They mostly went out of business or were bought by Verio.
Today, few ISP's actually have upstream bandwidth equal to what they've sold downstream. And it gets even more complicated when you consider that there are usually multiple routes out of an ISP, some of which can be easily overloaded, others less so, depending on where the traffic is destined.
The only worthwhile measure is a subjective one. Can you get 1.5Mb throughput on ANY site? On some sites but not others? Do you think you're going to get better service from somebody else? There's no exact answer as to whether you're getting your money's worth; experienced net admins have used a several connections over time, and usually know within a day or two whether they're on a good one or not.
If only all the other stupid fucks that post inane shit would do the same.
First of all, if you have *the whole T1* then you should have control of the CSU/DSU and the router at your end. Accept no substitutes. If the guy is, in fact, splitting a T1's worth of bandwidth off a DS-3 (say) then you should (at the least) have monitoring privileges on the router and DSU, either via web, SNMP, or telnet.
As mentioned earlier, the DSU should show that your connection to the line is using 24 x 64Kbps channels, for a total of 1.544Mb/sec (minus a few k for channel overhead gives you 1536). ALL of those channels should be coming out your end of the CSU/DSU.
If you have control of the router, then you might try (again, as mentioned earlier) ping-flooding the router at the other end of the T1 and checking how many packets/sec get through, then multiplying by your packetsize. That should give you a reasonably close answer.
If the guy isn't bullshitting you, then he should have no problem giving you read access to the SNMP MIB on the router and letting you watch traffic - if you can access this, you can run bandwidth monitors like MRTG.
Good luck.
A hero is someone who knows when to run away. I am a hero. -Trent the Uncatchable
Any T1 you get from anyone is going to be oversold. What matters is the speed specified in your SLA. If it says you're getting a 1.54MBPS T1, then that's the bandwidth they
have to provide to you. That doesn't mean though, that they have to have 1.54MBPS x the number of T1s they have sold worth of bandwidth, because not everyone is going to have thier T1 pegged 24/7. If your SLA specifies that you're going to get a certian speed, I wouldn't worry about it. Get the better deal, if they don't hold up thier end of the bargian, you get a refund...
Just set up a site that could somehow be submitted as a Slashdot story, wait for the Slashdot effect to bring wrath to your server, and look at your MRTG graphs while it lasts. :)
3 things:
1. tell your boss to fire you, and hire someone with clue.
2. you have a router, use it, dipshit.
3. see #1 and #2.
Jesus christ, do you think the router's there just to look at?
One of the interesting things I picked up on at my last job, is that service from the LEC tends to be better than other companies. For example, an Ameritech line is often faster than Broadwing (even though Broadwing is having Ameritech install the circuit anyway...). There are two potential reasons for this....Either they are screwing with their competitors lines, which wouldn't be a surprise, or they have better peering relationships with various backbones and other providers. The quality of peering relationships is important, and not something that is easy to determine.
I'm pretty sure the first rule is you do not talk about fight club.
While I have no advice to find your T1's speed, I have a suggestion to improve it. Take the hub that you and the other two occupants share, sell it, and spend the money plus whatever else it takes to buy a switch. Hubs devide up the bandwidth, so you're probably getting around 1/3 the speed you could be getting, switches don't. They only forward data to the appropiate computer/network, where as a hub broadcasts it to every device on all connected networks barring a switch/router somewhere down the line. Add a switch and you will see a great improvment.
Request: ECM unit, 1000 km fullerene cable, 1 tactical nuclear weapon. Reason: Birthday party for foreign dignitary.
Ahem. From this, it looks like you're really just buying Internet access with a "T1" rate. 900kbps is almost as good as you're going to get on a T1. Maybe upto 1.1Mbps or so, tops. You have to allow for protocol overhead, latency of all equipment between you and the "other side", and congestion that may or may not exist.
In my experience all vendors, big or small, 'oversell' themselves. Remember, all salesmen are liars, have no souls and will all burn in eternal hellfire... They will try to spin you a yarn to try and differentiate their products where there really are no differences. I would say that if you regularly see 900Kbytes/s and have uptime in excess of 99.99%/year, that you are doing well with your T-1 line (not considering the cost of your line).
Honestly I think unless you have access to some FTP server or something like that just on the other side of their DS-3 you will have no reliable way to tell for sure. Yes, you could run a bandwidth meter like DSL reports or something like that but unless there is no congestion between you and the server running the test you may not get accurate results. In otherwords, you may have slow link somewhere in between that would make it hard to check how fast your connection actually is. The only sure fire reliable way to test is to run a check from your site to something just on the other side of your provider(your providers provider) if you can find out where they buy their uplink from.
You could have a T1 while your provider has a T3 or something but maybe they sold 100 T1's with that T3 uplink. (a T3 is 40-some T1's) so you may never get close to T1 during peak traffic hours. (But it may appear you have a full T1 during off hours) Unless you show up and ask for a tour of their co-lo or office I think you'll just have to take their word for it that they own enough bandwidth to cover your needs.
Say they get their uplink from WorldCom, if you could find a server on the WorldCom/UUNet backbone where you could download a large file and check the speed during peak hours it might give you an idea. (Just incase it is WorldCom, http://www.sunfreeware.com has a mirror site on the UUNet backbone)
Good Luck...
Nail their upstream with pings... really big ones, down the pipe as fast as you can, right into the closest provider from them (their gateway on the other side). Then use (guess what) MRTG to meter the bandwidth. Reason for doing it this way is that almost any method that rates a download will come out on the conservative side because of network overhead, and you get to measure both upstream and downstream at once. Just play nice about how you this, massive ping floods can be taken the wrong way!
SIG: HUP
If you can get a smart bits and want to test a point to point connection its the best way. You can probably rent one from GE Capitol. They cost around 40k to buy so there not trivial but they are the best method of testing. It however requires access to both sides of the network to do properly.
http://smartbits.spirentcom.com/
--- Always remember. 99.36% of all statistics are inaccurate.
its not the answer, but for people who like a basic bandwidth meter for their system, go to the analogx site and get Netstat Live...
-- Betting on the survival of the media industry is a serious risk. I advise investing elsewhere.
I really can't believe this.
A few days ago I post a good story abount how the new terrorist database system keeps crashing, and get rejected:
2002-08-23 20:52:48 Terrorist database often crashes (articles,news) (rejected)
But somehow, this makes it onto slashdot...
On the other hand, I don't wanna get off topic so why don't you just do like everyone else, and get MRTG and set it to SNMP to your router and get the interface statistics directly.
"A terrorist is someone who has a bomb but doesn't have an air force." -William Blum
What commited information rate are you paying for? It's possible can get more effective sustained throughput with a fractional T
with a higher CIR than a full T with lower.
The full T will have higher burst speed.
Are you really inerested in some kind of bragging rights to say "yeah I have a T1 look at me" or are you more concerned with bandwidth. If it's thebandwidth it won't matter if its from a split DS-3 or a true T1 as long as your speed is what you need and expect. Besides you could get a DSL line running at 1.7mbs and it would be just as good considering a T1 runs at 1.5mbs anyway.
But I guess i'm just being logical here and forgetting the corporate thinking of:
more expensive = better
Ave Molech Setting
I run the network for the dorm here. I know that we have 1.544Mbps full duplex on both T1s because of...
P2P software. Yessir, these suckers are fully saturated at all times as the year goes on =)
Seriously though, the way I've tested is get a machine a few hops away, and start moving as many bits as you can. I use RRDtool to track everything, and it works quite excellently. I have multiple graphs, which collect data using SNMP directly from the routers...
matusa
A DS-3 is essentially a T-1 on steroids. Instead of only having 24 64k channels, a DS-3 has 672 64k channels.
So if the ISP is selling you 24 channels of the DS3 and calling it T1, then you're getting a T1. Quitcherbitchin.
Nathan
S.L.A. (Service Level Agreement)
If said small provider is telling the truth, then he won't have a problem signing one. I've found in my area that the big guys are the bullshiters when it comes to SLA's.
"Klaatu, verada, necktie!" -Ash
This even applies to phone lines.. That's why It's sometimes hard to get a call through on Mothers' day. The phone companies provision to handle 99%+ of the volume spikes, and mothers' day can consistently make it into that last percentile. Even so... handling 99% of the traffic spikes still comes to far less than one circuit for every two subscribers.
To test if your T1 is really a T1, you can try setting up for an FTP (or whatever) session with your buddies... Change things to make sure that you've each got the route to the other going through the first hop on your T1. As long as that router doesn't icmp-redirect the traffic, (in which case your measured bandwidth will be more in line with local ethernet traffic) you should be able to get a good estimate as to the raw bandwidth of your 'last mile'. (it'll actually be the lesser of your uplink speed and your downlink speed minus a little bit)
The next hop would be to set up a transfer with something inside of your ISP's primary network. (did they give you a web site on one of their systems, etc??). That'll allow you to test for local bandwidh bottlenecks and give you a theoretical maximum to the outside world.
The last link check would be to find a machine on a fast network that's not on your local ISP's but is (topologically) close. Try doing traceroutes to nearby universities.. See if you can find one that doesn't put you through 3 different ISPs. Then try and transfer data from/to them and see how fast it goes.
You'd be best to try connections to a few semi-local sites. Otherwise it'll be hard to tell if a low bandwidth reading is the fault of your ISP, or the server's ISP.
It's pretty much useless to check bandwidth to random (distant) sites. Once you get a site that's a reasonable ways away (topologically or geographically), then you run into the vagraties of internet traffic (see the article earlier this week about 'net quakes)..
BTW: When I speak of being topologically close, this is different from geographically close. I remember one case where getting a packet to a machine not more than 100 feet away (but on another ISP) sent traffic from Vancouver, down the coast to Silicon Valley and back. Needless to say, ping times stank. In that case we were geographically close, but topographically distant.
THere would be two times to test these transfer speeds: Low time (e.g. 4am) and prime time (Last time I peeled apart ISP traffic stats, traffic peaks were around 8-9pm for home traffic and about 4PM and 9AM for commercial traffic)
Free Software: Like love, it grows best when given away.
I thought I had a T1 but it kept asking if I had seen John Conner, turns out I got a T1000! Word to the wise NEVER shop with Cyberdine!
-Jason
I worked for a small ISP (40,000 subscribers) for 5+ years. The best method to tell if it is over subsrcibed to to run bandwidth tests at various times during the day and night and compare. Of course you need to also look at the price that you are currently paying and compare that with what everyone else is quoting.
If you are happy with the price that you are paying and the service/bandwidth that you are getting then stay where you are. If you are unhappy then go shopping.
Even the national providers oversell there Internet service. They just do it on a different scale.
There seems to be some confusion between what a T1 is, and the data rate a T1 can support, as well as how bandwidth is marketed and sold, and wether or not the online "speedtests" mean anything.
... or raise several billion in VC money to build your own Internet to all the sites you want super-fast connections to.
First off, excluding some very odd and sparse cases, from a telco "loop" perspective (the wire and associated hardware involved in providing a T1 from you to the telco's office or remote), a T1 is a T1, regardless of it being frame relay, fractional, point to point, or otherwise. Basically, 1.544Mbps raw data rate. The only thing the telco can tell you is that it is working within acceptable limits, with an acceptably low BER (Bit Error Rate). Nothing more. If it's a non frame relay dedicated T1 (aka point-to-point or "nailed up"), you should see something like 160kBps on a single transfer over an otherwise idle circuit. Frame relay is a totally different ballgame - you run into the circuit's "CIR" - Committed Information Rate (or also referred to as Certified Information Rate on occasion). Basically, it works like this - even though your local loop (wire from you to the telco) supports T1 rates, they're only provisioning your chunk of the frame relay "cloud" to support the bandwidth you purchase. Depending on the provider, it could be a "hard" limit (ie - you'll never get more than that), or a "soft" limit, meaning that you may get more speed if there is sufficient packet bandwidth left, but when it's busy, you'll get choked down to your CIR.
Second immutable fact of the Internet - providers (particularly tier 2 non-backbone providers) will always oversell their bandwidth. Think about it - how do they make any money at it otherwise? It's their bandwidth "sold" to bandwidth "available" ratio that tells you anything about the quality of service you may expect to see from them. It can be virtually anything really. It's similar to the subscriber to modem ratio dialup ISP's keep, or subscriber to bandwidth broadband ISP's keep.I'm not sure what the averages are anymore with the changing scene of broadband right now.
As for testing your speed with one of the online "speed test" sites, take the results with a grain of salt - a very BIG grain. It's only benchmarking the ability for you to transfer data between your location on the Internet to their location on the Internet. Nothing more, nothing less. More oft than not, their results are less than what your circuit (or broadband connection) are configured for, and sometimes, dramatically less. I've seen 256k DSL connections that "felt" as they should speed wise, bench in at dialup speeds using those sites because of bottlenecks beyond their or their provider's control.
A better test of bandwidth, and possibly more importantly connectivity, is to do some basic homework. Ask your provider to provide some traceroute and ping data to some common sites potentially used in your day to day activity - taken at different times of the day, preferrably at the times you're most interested in. Or better yet, ask them if they would let you do the tests yourself from their facility. Choose some sites you frequently visit, business or pleasure, and trace their progress theough the Net. Fewer hops = better speed and reliability overall.
I may catch a little hell for this, but I've found it to be a good general benchmark - download something that is known to be cached at at Akamai cache farm. NAI virus updates used to be a good test. Akamai is generally connected very well, in strategic points through the Internet, and provides at least a benchmark to go from. Downloading from Microsoft is sometimes a decent test, sometimes not, depending on what's downloaded and when. And, no, that wasn't meant to be a troll or flamebait, just a basic fact really.
In all, if it generally looks right, feels right, and provides stable, repeatable performance during the times you need it, it's probably fine. There is no good, clear, black and white method of determining if your "T1" is a "T1".
They may just be shooting you a good deal because they have plenty of bandwidth, and sales prospects are limited because of the economy. Who knows. Just do your homework, limit your exposure to marketdroids and the resultant weasel-words, and go from there.
Remember, for all intents and purposes, the Internet is basically the data version of the public road and highway system. You'll find construction, detours, and bottlenecks regardless of how many 12 lane super-highways exist. You just need to live with it and work within those limits as best you can...
Brad
All bandwidth providers oversell thier available capacity. Almost no one (legitimate) uses all of thier available bandwidth all the time, so usually everything runs together fine. They have to do this to comepete with everyone else. A T1 isn't really a big connection any more and you're probably not going to be an account they bend over backwards for. If you really want to be on a network that isn't oversold, you're going to pay 5 times as much. Worry about response (ping, number of hops in traceroute, etc) times to AT&T @Home customers and DSL customers. At the t1 level if those response times are good, you are doing great. When you move into a DS3 and up connections, the whole sales game changes again.
I have a remote-administered server in a colo in Mountain View, CA. I routinely have transfer rates of 2200 Mbit, for example, when updating via the Red Hat Network.
/dev/null on the system being tested, and watched the graph with MRTG.
Yes, that's about 15 T-1 lines worth of bandwidth - my 1.5 Mb DSL line is always saturated when loading stuff down from it.
So, I wrote a quick PHP script to take a 100k file off the disk, and hurl it at you repeatedly 1000 times, with a mime-type of application/octet-stream. That's a 100 MB file being downloaded by HTTP.
Then, I hit the script with lynx -dump >
In my case, we did get effectively 1.48 Mbit on the T-1 line being tested, which is close enough to satisfy my curiousity.
I'd be happy to perform this test for you... BTW - if you are in North California, I recommend o1.com as an excellent provider!
I have no problem with your religion until you decide it's reason to deprive others of the truth.
Your subject is great, but your body is lacking. With a working T1 and good network, it's quite easy to get upwards of 1.4Mbps of data throughput.
T1 consists of 24 channels.. DS3 consists of multiple channels as well. There is nothing wrong with taking a channel of a DS3 and using it as a DS1, in fact that is typical.
That provider declared bankruptcy (doesn't narrow the field of suspects too much I'm afraid) and left California. We now have a T1 split into half voice and half data - ie. half the bandwidth we did before - and everyone is amazed at how fast things became. The real difference is that the round-trip latency is generally more like 20-40ms or an order of magnitude faster than with the previous carrier.
Note: we are with a "smaller company" and are quite happy with the service. I've "been serviced" by the big unnamed telco (Should Be Castrated...) and it has always been terrible. Terrible for home voice, terrible for home DSL, terrible for business voice and terrible for business data. The big guys are doing what they always do - spreading FUD instead of providing good service.
~~~~~~~
"You are not remembered for doing what is expected of you." - Atul Chitnis
Guilty as charged. But with an explanation... I was just assuming he's been getting his 900kbps with remote bandwidth testers (located who knows where) on "untweaked" PC's.
There is many hops between you and the server/clients elsewhere on the net with who you want to transfer data. You want a "T1", and want to know the difference between a bad and a good T1.
I am no network ingeneer or have the pretention to know anything about this but here is how I see it:
A T1 is the speed of a link. When jumping a hop, that measure is not anymore a good one. The next hop, out of your provider, will probably not have only you for client. Then it depends of the speed of its external link and the number of clients * the utilisation of theses clients.
My cablemodem provider does not have an external link to the net fast enough to support the number of its clients * the speed I can get.
And where is the data routed when it leaves your ISP? Maybe it can be a factor...
The ultimate test would be to use few servers, placed in areas where you will often transfer data.
Then transfer many packets of a fixed length, and check on the other side the latency and the throughput (with each one). Repeat again every 2 hours, everyday, for x days, using both T1 providers. Compare the results... Depending of the quality of the ISP external link (and who's at the other end), and the traffic generated by the other clients, you will get different results...
Good luck in your quest
dslreports.com
FoundNews.com - get paid to blog.,
If you want to test your T1, unplug the token-ring cable and count the number of tokens that fall out.
Just remember to put them back when you're done counting.Do you like my new sig?
You should definitely consider replacing that hub that you and the other tenants use to go out on the T-1 with a switch. Your tests aren't really going to tell you much when you have a hub in the picture because you are creating a potential bottleneck before you even hit the T-1.
-- Just my $0.02 worth...
I have access to a T1 line after hours THROUGH A 802.11B WIRELESS LINK and I consistently score 1200-1300 bps with the dslreports test. I think you may be getting screwed.
"Eve of Destruction", it's not just for old hippies anymore...
I agree, about the title at least. For a minute fraction of the cost of a T1 you can get DSL with high upstream, even with an SLA. For example, take a look at a local ISP: http://www.istop.com
iStop offers 2.3mbit SDSL with an SLA for 205$ canadian. A T1 costs about 1000$ canadian, and only does 1.5mbit. Need I say more? Even the 800kbit upstream for 85$ is amazing compared to a T1.
Now, I'm not just pimping my ISP. Many others around here have similar plans (I'm just most framiliar with my ISP's plans). T1s are old, outdated, and overpriced.
Regards, Guspaz.
Is it the name you're looking for or specific bandwidth? You really should just ask for a bandwidth guarantee. He may not be overselling his bandwidth to you, but to his next 800 customers. What if your performance turns to crap in a few months and you're stuck holding a contract and 5 K/sec d/l speeds (and probably a pink slip)?
Furthermore, we split the T1 out over a hub with two other tenants in the building. I'm coming through from behind that hub.
Hubs? You actually use hubs? No wonder it seems slow.
Here is a Google cache of the difference between hubs and switchs (basic).
If you're in charge of the network, you need to take some courses. No offence.
I'm a network performance engineer, and have tried most of the tools out there to test available bandwidth between two locations.
The short answer is - you have to do file transfers. Use HTTP instead of FTP for the best results (I like wget personally). Suck a large (10MB or more) file from the ISP that your provider gets his DS3 from. Repeat that a couple times an hour for 24 hours (wget via cron is an easy way). Save the throughput results and see if they vary during peak hours. If you're getting over 150KB/sec (1200kb/sec) you're fine.
All the artificial tools in the world won't tell you the real answer, only real transfers provide good results. And yes, the other people you share the T1 with can contaminate these results, but probably will not unless they are doing the same testing or hosting a large download site.
- j
You can have a fast pipe, but one of the advantages to a t1 is that it's a wider pipe... who cares if you get 200kb/s downloads if only one user can sustain that. When I start talking about t1's, I care more about how much data I can push through, vs. how fast I can sustain a single download. Your response is somewhat shortsighted, I think
My sig sucks.
One thing that always bugged me was how the Time Domain Multiplex (TDM) formats naming doesn't make a lot of sense. A DS-3 carries 28 T-1's. A DS-3 is sometimes known as a T-3, and maps closely onto an OC-1. The OC naming makes a bit more sense, in that OC-3 is three times the capacity of an OC-1, OC-12 is twelve times, etc.
If the ISP has more than 28 T-1 customers and a single DS-3 (or OC-1) to its provider, then you are all sharing that bandwidth. But if you think about it, at some point you are going to have to share comm. links somewhere upstream, so the fact that an ISP has only a DS-3 is not in itself a concern.
About the only thing you can do is keep an eye on the performance that you get, and ensure that you (1) can achieve peak bandwidth of 1.544 Mbps sometimes, and (2) that your average bandwidth isn't too low. Ask your ISP what average bandwidth you can expect to its provider, and use that as a benchmark. If your average gets too low, then you can complain.
Anybody want a peanut?
A full blown T1 is 192 thousand BYTES per second (192kBps, 192000 not 192 * 1024). Go to your favorite high bandwidth FTP site and start a MASSIVE download during prime time and see what type of through put you can get.
I run a medium sized ISP and we 'oversell' our bandwidth but we still guarantee the full T1 through our network to our upstream providers. It really isn't that hard. I may sell 10 bits for every one I buy but only 1 out of every 20 customers needs a bit at the same time. It all works out.
The big guys are over committing the EXACT same way but they are 'the backbone' according to their sals droids so in theory you have a T1 to the backbone. Problem is what happens if all your data is going to another backbone (i.e. UUNE to Sprint) How well is the big guy going to hand the data off?
All told a T1 is a T1 from most decent providers the deciding factor is the service you get from the people
All ISPs even the phone company over sell their down stream to upstream. 4:1 is a common ration for business level down stream bandwidth to upstream bandwitdth.* I know of one ISP here that keeps this ratio, and generally uses about 50% of their upstream bandwidth. They are likely to have burstable upstream connections, just in case it gets extra busy.
*Incidentally, consumer level stuff has higher ratios that that, I have heard of dial-up places with ratios closer to 200:1. I doubt broadband ISPs can go much higher than 10:1, but that is just a guess. That ratio is one of the best indicators of service quality, but I doubt you can find a service provier that will tell you theirs.
Degaussing scares the bad magnetism out of the monitor and fills it with good karma.
Most of those friggn things measure the "time it takes to download a 1 MB image file"
;)
What they don't tell you is that its mostly wholly dependent on the rendering speed of your computer!
Two machines, side by side with different processors, browser versions, video cards, drivers will give wildly differing opinions of the true "speed".
Unless you're doing a lot of hosting on your end of the T1, or you're doing QoS based services (like VoIP) -- go with the local ISP -- the money you spend will be returned to your community and you'll get better support. Not to mention, most of the smaller ISP's do other types of consulting, so you almost always get more bang for the buck.
But take everything I say with a grain of salt, since I own a small local ISP -- not in your area
Old age and treachery almost always overcome youth and skill.
Do not order the T1 unless the CSU/DSU (Channel Service Unit / Data Service Unit) is ON YOUR PROPERTY. This usually means you have to buy one, which is usually in the range of $700, give or take a few hundred. I wouldn't trust an off-site CSU/DSU.
At this point, you know for sure that you're getting the 24 channels @ 64k. The only thing you have to worry about now is if the ISP's backbone can support you, which is going to be a problems with EVERYONE at the ISP, not just the T1. For example, I have 1.5mbit DSL. Not the same thing as a T1, but we're both tied into the same aggregation circuit. In that case, do NOT trust the speed test from the ISP's FTP since you are going to get your 1.5mbit from the ISP, that's not an issue. The issue is getting 1.5mbit from an FTP CLOSE to the ISP, within a few hops.
So again, you're gonna get 1.5 to the ISP aggregation circuit. Easy. Getting 1.5 OUT of the ISP is what you're concerned about. Although as soon as you hit the ATM or frame relay cloud it's smooth sailing (or OCx, depending on the ISP).
Or, screw the T1 and get frame relay yourself with the appropriate CIR.
Most folk'll never lose a toe, and then again some folk'll...
As we all know that somewhere down the line, someone's going to be oversold, you have to ask yourself if you are getting what you want for the money you pay.
:)
If you're getting 150-170KBs on these tests, but the price and service you receive is excellent, does it matter?
If you are getting 300-900KBps, you're probably getting more than a T1... keep it quiet.
2002-08-23 20:52:48 Terrorist database often crashes (articles,news) (rejected)
But somehow, this makes it onto slashdot...
Like anybody in the know actually expected the terrorist database to be stable? It's a political playtoy! It might work one day but, in the meantime, everybody still has the database that they're used to using.
At least this article has people thinking and talking. Methods of testing bandwidth are the kinds of things that have true geeks sitting up at night thinking.
OS Software is like love: The best way to make it grow is to give it away.
idiots! shut up idiots!
read a goddamn book, learn the OSI model, you have just sucessfully proved that there is such a thing as a stupid question.
Couple of things, first, all ISPs oversell bandwidth. I am assuming the 'larger' guys meant that the local ISP likely only has a single DS-3 to his upstream provider. Depending on how small he is, this may not be an issue (/all/ providers oversell at some ratio, it just depends on what that ratio is that determines if the provider sucks or not) depending on how many other customers he has, what types of circuits your providers customers' have, and what type of customer they are (business vs. residential), etc. Obviously there are other factors too.. if your provider suddenly grows and doesn't increase its upstream capacity, thats an issue.
/most/ circuit IDs from /most/ telcos is something like XX.AAAA where AAAA is frequently what your concerned about to determine the circuit type, but the format of the CID depends on your telco (there are many guides out on the internet at decyphering these to determine what type of service they are, or you could call your telco and they could tell you.. maybe.. if your good.. heh)). The CID should be on the smartjack..
/can/ be, without having to deal with other provider's networks being congested, etc (the traffic in this case will be local to your provider's network, so if thats congested then um.. that sucks).. You should get somewhere around ~192KB/s.. if its slightly less don't worry about it, there is some overhead involved, etc.. When you do this be sure you only have /one/ machine connected to test (or you can verify there is nothing else that is generating traffic that is going over that circuit, etc... don't assume.. check. (there are many, many, many tools to do this..) to see whats hitting the ethernet interface of your router (its a lot easier to check if you have access to your router, as you can just do a show int on a cisco to get traffic statistics, etc).
Now, on to determine if you have a 'real' T1. Many providers tend to sell frame relay service as a 'T1'. While its true that the circuit itself from the telco to your place of business is a T1 (unless you have say 56k DDS service, etc), after your traffic hits the telco's switch, it transverses their frame relay network, and eventually gets sent to the frame host on your provider's network. This can suck for many reasons, however the biggest one is that the provider can get away with purchasing a CIR that is less than 1.5 mbps (like 768kbps), and just have it be able to burst up to 1.5. This can suck a whole lot if the telco's frame network is congested, and you can never burst, and/or you constantly want to use more than the CIR etc..
You can tell what kind of circuit it is by quite a few ways, if its installed and you have access to the router you can simply check the encapsulation on the T1 (if its set to frame relay, its a frame circuit, if its set to PPP or HDLC its a point to point circuit), you can also tell by the format of the circuit id (the first part of
If you have a frame circuit they should be charging you quite a bit less than a point to point T1, especially if the CIR is low (the lower the CIR the less expensive generally). Point to point T1s are preferible in almost all cases unless your worried about cost.
So now that you've determined the type of circuit you have, you can check what speeds your getting. Its usually best to do as other people have mentioned and download (and UPLOAD too.. you should check both speeds) from an FTP server on the provider's network. This will give you the most accurate picture of how fast your connection
You should also do a separate test to a major site.. You could download the 1.4.0 Java SDK from Sun for instance (that should give you a decent speed). Don't worry if this is somewhat lower than 192KB/s, as that can be caused by congestion in a network inbetween you and sun that has nothing to do with your provider, etc.. If its consideribly lower than that speed consistantly (and other sites with bandwidth to spare yield the same results), then I would contact your provider about it. It could be that your provider doesnt have enough upstream capacity, or about a billion other things, but they might be able to tell you any known issues, or that the problem isn't them and/or tell you the current utilization levels of their upstream circuits.. heh
"real" T1/DS1 uses some low level framing bits (ESF)in addition to packet and tcp overhead, but now days, that's not the major concern as there's a bunch of "cheap" T1 resellers that have only about 3 to 6 T1 equivalent bandwidth connected to major NAPs but oversell this bandwidth to other folks... (yes there are really people who sell 20-25 people T1 bandwidth even though they only have 6 T1 bandwidth to the rest of the internet)
Even T-3/DS-3 (43Mb/s) connectivity isn't much if they really oversell things or if they have cheap non-linespeed routers or bad peering agreements...
I like the approach mentioned earlier in the thread about initiating remote loopback (assuming you have a DSU/CSU and can do it). You should be able to write some simple perl that opens and writes a file of known size "across" the loopback connection to a program listening on a socket on the other side and calclating the bandwidth based on the file size / duration. That would really only measure the transport layer payload capacity, but I'm sure it would be pretty close to accurate. I've done this sort of thing before (with wireless PPP connections) with excellent results.
This idea was invented by Shampoo.
testing your outbound traffic is going to be a bit harder (mostly a problem with finding people who will take great gobs of your data).
P2P networks with a juicy set of files to transfer are a pretty good way of doing that.
If you can watch the raw traffic as it goes to your DSU, then you should even be able to factor in the bandwidth that your building-mates are using, too.
Free Software: Like love, it grows best when given away.
How much does the price for that 2.3mbit SDSL change if I wanted to attach a mini-ISP to it?
:-)
For example, lets say I want to run an internet cafe & a (small) co-lo server farm off it, using (lets say) 1/4 to 1/2 a class C -- how much would it increase the price (just a rough guess would be nice)?
Just wondering 'cause I'd like to settle the debate with some numbers. Not to mention I'd like to know for my own company's future.
Thanks!
If you could be told what you can see or read, then it follows that you could be told what to say or think - BoC
But I digress.
1.) Find out exactly what sort of bandwidth he has to the backbone provider(s) he is using.
2.) Check out THE BACKBONE PROVIDERS he is using. If he's a straight UUNET shop, then what's going to happen as the overworked staff at UUNET continues to shrink with the next round of layoffs? If not UUNET, then how good and stable are his backbone? His backbone provider could be some sort of half-assed operation itself.
3.)How many IP addresses is he advertising through that DS-3 (or whatever) he has? Does that sound reasonable?
4.)Be realistic. You can have cheap, you can have good. Pick one. Most internet connections are oversold. That's the way it is. If your ISP has 400 MB of customers connected to the 45 MB pipe, you're doing pretty good. If you demand a 1:1 ratio, then buy from the major backbone carriers and stop fooling around with a small guy.
5.) Check out the BIG guys. I know of one VERY BIG (and non-bankrupt) backbone provider offering T-1s at $1000 pretty much anywhere in the US. You want to spend significantly less than that? Well, you get what you pay for.
6.) SLA's can mean something if the carrier has some integrity. If the SLA is intended to be an accurate prediction of the service you get, then great. On the other hand, lots of carriers view SLAs like insurance policies. They balance the money they pay out when they get caught with the money they gain by bamboozling gullible geeks who just compare the numbers. Some SLAs are honest predictions of performance, some are bullshit sales hype. Anyone who promises 100% uptime fits the latter category.
7.) Finally, most of the tests suggested in most of the postings on this topic are nothing but bullshit. The methods outlined here are a lot better than FTP'ing a file 100 times and running an average.
I work and design NOC for CLECs, there are several type of T1s first type is Standard T1 where you get all 24/64k channes, 2nd type Fractional T1 this is where you get all 24 chanels but you are caped at band with, 3rd(and these you have to watch out for cause they get expensive this is also what the big guys like to sell) Fractional Burstable T1, This is a soft cap on bandwith but will meter any thing you go over. 4th is a split data voice T1 dont see these much any more. Any way if you have a T1 it is proabally a T1 however you have to watch out for how they charge you band with and things like that. Another issue to wacth out for is local loop fee. And IP allocation fees.
Linux modi 2.6.26-2-parisc
I dont see how that is considered informative considering it doesnt provide information on whether something is a true t1 or just as fast as one. My dorm connection is listed as a excellent t1 connection but is not a true t1 with dedicated bandwidth.
ok assuming you have access to the router on your end you should be able to do a ping with large packets. Assuming nothing else is going on you should be able to send approximately 1.536Mbits of traffic (minus overhead) every second. So turn off your outbound routes and do your ping flood to establisha baseline for max return. Then trace out from your site to some nice high bandwidth site on the net and repeat this test at every hop. You should be able to determine which IP's are within your hosting providers network. You should get full T1 rate from any of these IP's. Once you figure out the IP's of the other end of your upstream providers network you'll be able to accurately figure out how oversold your provider is. Just do your ping flood for 24 hours and see if you always get near max or if it varies with time of day. If it gets slow at peak times you can gauge how oversold your ISP is. This will get more complicated if you have a large upstream provider with multiple links who is using BGP since you'll have to take that all into account. Perhaps if you can determine how many outbound links your upstream has you can do this test with a ping flood through all the links that way an individual link saturating wont skew the results. all this other talk about watching mrtg or cricket for general use is bullshit - you wont be able to seperate out the internet general problems from the oversold problem. Take a scientific approach to things rather than just looking at some pretty graphs.
300-900kbps is most definitely not a real T1. He has likely oversold his upstream. We have a real qwest T1 at work, and most file downloads are in the 130-170KBytes/s range.
Extra hops add latency too. In any case, the salesman is probably mostly correct.
But, it still may be a better deal for you to go with the local guy. You need to ask yourself, are you getting the T1 and plan to use 100% of it bandwidth wise, or are you getting it for the freedom that comes with a T1? If it is more the latter, the local ISP might be a good deal.
I've had enough abrasive sigs. Kittens are cute and fuzzy.
Just put up a website about how you hacked a [insert crazy tech device here] and managed to get linux running on it. Have Slashdot run a story about it on the homepage, and if you can handle the rush of traffic, then who cares if you really have a T1?
Dude...you almost had it. If you had included a link to a webserver running on your network in your post, half a million /. drones would have clicked it.
Then all you have to do is monitor the traffic between your gateway and your webserver. Presto! Instant T1 validation.
"Don't blame me, I voted for Kodos!"
Ask him the "Certified Internet Rate". A full T-1 line delivers 1.5 MB/S through 24 separate channels, each channel is responsible for 128K. For instance, the T-1 where I work is separated into 12 circuits for internet (768kbps) and 12 for voice. My CIR is 700 kbps down, 256 up. That means I can download something at a guaranteed minimum of 700 kbps, and I can upload something at a guaranteed 256 kbps.
He cannot, by law, lie about the CIR. If he does, the company he works for gets major fragmentation from the shit that hits the fan. I can sell T-1 service all day long, but if all 24 channels only amount to a CIR of 1 MBPS, I'm hardly competitive with companies who are selling my T-1 lines with 1.4 Mbps CIR.
BTW, why are you getting a T-1 line? Are you splitting it for both voice and data, or is it pure data. A number of clients I've consulted for in the past really only needed a cable or dsl connection for their uses (basically, web collaboration and surfing).
There's no way to answer the question you asked. But, maybe you should be asking a different question: Am I getting what I'm paying for?
/. earlier in the thread, and see if you're satisfied with the results.
There's no way to tell from the outside how much bandwidth the provider has sold, and they're not going to tell you. All you can do is, use any of the lovely tools touted by the great geeks of
Probably best way to do it is, find a few sites that you can get high throughput to (>1 Mbps) on off-peak times. Monitor those connections at peak and off-peak times for a week or so. If there's a big difference for *all* the sites, chances are your ISP is not giving you a big enough slice of the pie. If it drops for some and not others, you're looking at other parts of the internet getting bogged down. Even this is far from a perfect test, but it can rule out ISP bottleneck.
Don't you wish your girlfriend was a geek like me?
Do a search for "xxx" sort by availability and download 30 or 40 at the same time.
Conversely you could put up a bunch of mp3's with popular song titles and the outgoing will have the same effect.
this is a GREAT way to test line rate, and has worked well for me for quite a while!
I dont know if this is any use, but try Tmetrico pic_id=87
http://freshmeat.net/projects/tmetric/?t
sor of based on pathchar, its a bandwidth discovery tool which from memory gives a traceroute-like output, including bandwidth estimates from jump to jump
We had a T1 from PacBell, on the central coast of California. DSLReports.com tests worked fine, read 1400s usually. Note that that is data bandwidth and so doesn't count tcp headers, etc. DSLReports claims that their tests are usually accurate to 5mbps or so. If you're getting 300-900 from reputable sites... you have middling cable speeds. You should be reading at 1200 at the very LEAST (1544 is of course T1's rated speed)
Save time now so you can waste it later
2. Obtain naked pictures of at least 2 of the slashdot editors or worst, them using a windows ce device.
3. Get kuro5shin story to point at your new web server.
4. Reap money from fire insurance and buy an oc-3.
An Education is the Font of All Liberty
If a PC requires tuning to sustain 1.5Mbps you have major problems, or you need to upgrade your 286. 1.5Mbps is extremely slow in computer terms. The PCI bus has over 50 times that capacity.
I've had enough abrasive sigs. Kittens are cute and fuzzy.
Check DSLReports.com, they're one of the best-rated DSL services in the country, and their T1 service is great too. Give them a call, (800)556-5829.
Moo
Well, here's a quick test to see if the line is over-utilized. Ping.
Trying transfer speeds from other sites simply doesn't work. Not only are you dealing with your own latency, but with latency of every connection between you and where you're going.
A quick traceroute would give you what you're looking for. Run the traceroute to a known T1 customer on their network. The last few hops will go something like this:
5) Some router on the Teir 1 provider.
4) Connection between the Teir 1 provider and their equipment.
3) Their equipment (router) going towards the customer.
2) The customers router.
1) The customer's host machine.
Stop reading at the first place you see bad numbers (say above 150ms), or "*". Anything after that is caused by the problems previous.
If you see shitty times before 5, it could be your provider or theirs.
If you see shitty times at 5, that's their providers problem, don't use them.
If you see shitty times at 4, they may be overextending their available bandwidth or a line problem. Don't use them.
If you see shitty times at 3, there's conjection on their LAN. Don't use them.
If you see shitty times at 2, that particular T1 may be overextended or there's a line problem. This isn't bad, it's that customers problem.
If you see shitty times at 1, there's problems on that customers LAN.
Everyone oversells their bandwidth. If I have 45Mb/s to resell, I may have 100 T1 customers who never exceed 100Kb/s each. There's no reason if I (as the bandwidth provider) can't sell more T1's on that connection, if I'm only utilizing 10Mb/s now.. It's my responsibility as the bandwidth provider to be sure that we never exceed a limit (my provider is 40% utilization).
It doesn't really matter if it's a huge provider, or a local provider, shit happens. Some huge providers have really shitty bandwidth. They'll have an OC3 in a city, and sell literally thousands of T1's on it. At peak traffic times, you'll see tremendous packet loss (>50%) and huge latency (>500ms).
I've personally flooded a few T1's I'm responsible for, just trying to transfer too much at the same time to a place with better connectivity (Gigabit fiber). When that happens, I've overused the bandwidth, and you'll see packet loss and latency.
Serious? Seriousness is well above my pay grade.
If you have a T1 directly to his NOC and there is no contention on the line you should get approximately 180 - 190 kiloBYTES per second.
Have the guy on the other end setup an FTP server with a 50 meg file on it. And then just loop wget to download it over and over. You should get the exact same speed everytime no matter what time of day. Unless you're sharing it with someone else. Or unless he's overselling his pipe.
And definitely setup MRTG to monitor it.
Hubs devide up the bandwidth, so you're probably getting around 1/3 the speed you could be getting, switches don't.
Uhmmm... No. It's trivial for a gateway router to ignore traffic that is not destined for it. In that environment size the bottleneck will be the Internet connection, not performance loss caused by excessive traffic at the router's LAN side.
But, your point is not completely flawed. A switch sure as hell would improve security! Actually, there should be three firewalls behind this gateway, you're putting a ton of faith in your neighbors.
FWIW
As many others pointed out you will get a full T1 to the ISP. Where it goes from there is the bottleneck. You should find out how many hops it is from the ISP to the local backbone provider (SPRINT, UUNET, whatever...). The fewer hops the better. One hop away on a 110% DS3 can be better than 10 hops away on a 50% DS3 as each hop is a potential bottleneck. Good luck.
Karma: Positive. Mostly affected by the lack of a karma joke in your sig.
You have a TRUE T1 if you have.. a T1!
If you have constant 1.5Mbps between you and your ISP, tha'ts a T1.
And that's ALL A t1 is.
That's it.
It's just a line.
How much upstream bandwidth your ISP has is another issue. How much upstream bandwidth THEIR isp has is another issue as well, and so on, and soforth.
The best way to tell is to see what they will put down in writing.
Best thing is to ask for certain maximum latencies to certain points outside their network... that's the best bet..
I believe you wanted to test out for sure to see how much bandwidth you were realizing at YOUR SITE.
I will predicate the rest of my post by saying my method might sound a bit amateurish but it will work and i have used this.
I've found the best way to do this is actually go to a windows box and use this tool called FlashGet, load it up with downloads (from various sites, for load balancing) www.easynews.com can take allot of abuse if you have an account, and then download. You should probably test several times during the day. Perhaps once before business hours, immediately before lunch-hour, and late night after hours, to see what kind of bandwidth you can realize depending on the load. (If he's not overselling then you should find that the bandwidth you're getting is reasonably consistent)
The reason I suggest FlashGet for WinTel is basically it's a pretty good download manager that has a ton of options which make it easy for these kinds of tests. I've gotten this combination to saturate a T-1 before, so perhaps you should let people know you are about to "test" out some stuff that might knock them off for a few minutes.
Now if you don't have some kind of third party tool measuring your bandwidth, you'll have to sit around watching the bottom "status bar" looking at the bandwidth measurements, but like I said this is a pretty quick and easy (not to mention cheap) way to test out your bandwidth.
You'll get a full T1 to the ISP, but you want to know if you'll be getting 1.5 mbit/s from the ISP. Perhaps so, perhaps not. There is a good bet that during the busy times, you won't be getting a full T1, but you might. Since this guy isn't a Tier 1 ISP, chance are he's over selling it to make a profit, which isn't bad, since he's also selling you a cheaper link anyway. How to verify this? Run MRTG, or try to saturate this with a UDP bandwidth generating tool like IPerf.
Put the remote CSU/DSU in loopback mode, use a line tester on your end. It should tell you 1544000 bits/sec exactly. Or am I misunderstanding the question?
Try putting AutoNOC on your router
(http://www.AutoNOC.com).
It's commercial but comes with 30 day fully
functional free eval.
Aaargh. Serves me right for being lazy in my explanation...
What I was getting at is that if you just take a default configuration of your networking, and try to do bandwidth measurements with a remote "bandwidth tester" over a "public" network that has all sorts of hops between you and the test-server, you're likely to get not-so-good performance, and that 900Kbps is not a terribly bad number. As I said before, it depends partly on what doodads might be between the testing PC and the server (inclusive of the equipment on both ends). Poorly configured client PC's over a high-latency network can take a 1.5 Mbps connection and make it look like 400 Kbps...
You can, of course, run a test on a quiet and private network, with all the networking parameters tuned to optimal to get full utilization of your bandwidth.
This touches on some of the issues.
*sigh*
I worked with local ISP setting them up and I have seen hundreds of business plans and network designs and if any local ISP wants to stay in business then they will have to oversubscribe their bandwidth. It is simple math/economics. If they buy a T-3 (45mbps) for $20,000/mo then how many T-1s for $500/mo do they have to sell just to break even on their bandwidth charges?
Also, keep in mind that just because the local will give you a full 1.544 mbps now doesn't mean that he won't later oversubscribe his capacity.
Depending on what you are doing, latency is also an issue with locals. Find out if they are getting their bandwidth from a backbone provider.
Those who test by FTPing large files and watching the transfer rate, should understand these limitations (kindly explained to me by J.Spencer Love).
I had a similar problem trying to host a large-bandwidth video clip. It turned out the bandwidth of my 10Mbps line did not saturate at all (in fact, it was utilized at mere 5%), so neither did the trans-Atlantic connection. The bottleneck was the internal buffer in client and server software.
This also means you may not need that much bandwidth to push the speed of your FTP/TCP-based tasks to its limit.
17779 eligible voters in a district, 17779 'vote' as one. This is Russia.
Does anyone know of a company that has a faster network than Savvis (savvis.net)?
They've got SLA agreements that I have never seen beaten for on network latency.
The man who trades freedom for security does not deserve nor will he ever receive either. - Benjamin Franklin
Write up a short story that says your company uses Linux because they hate Microsoft. Then, place that story as an HTML page on a a webserver on that T1. Submit the story to Slashdot. (don't worry, it'll get posted.)
If you're not getting enough hits to it from Slashot, put a small Flash animation on the page. This will infuriate the users, they'll bitch about it and cause more people to hit your site.
If your server catches fire, then your T1 is up to speed.
Good link.
Actually, I sell electrical test and measurement equipment, and we've just picked up a new tester designed for just that. It's called the "Interrogator". Works very slick; push a button, and the automatic test will verify the number of active lines, etc. Manual tests are also available. Check it out at your leisure (http://www.megger.com) Here's the direct link. http://www.megger.com/us/products/ProductDetailsBy SubGroup.asp?ProductGroupCode=T102&ProductSubGroup Code=T102&ProductGroup=T1%20Testers&BusinessSector =T&BusinessSectorName=Telecommunications%20/%20Dat a%20Communications
Have a look, and feel free to give me a ring if you want a ballpark price, or if you want to try one out.
Regards,
Andrew Lamore
Electro-Meters
Inside Sales Representative
(800)617-3413 Ext. 206
alamore@electro-meters.com
Certainly, Qwest and Worldcom come to mind... but then again, these 2 are probably the 2 companies you want to get connectivity from if you can. Weird.
Also, if you live anywhere near a city, you shouldn't be paying more than $1000 per month TOTAL for a T-1 nowadays. (including your local loop)
Further, even if you get a T-1 from a big carrier, they will probably just be breaking you off a T from a DS3 anyways... so screw it... go BIG... Get so OC connectivity.
--
Custom Computers for Discerning Tastes
ask them straight up whether they are delivering 24 channels of DS0 64k lines. If they are only delivering 56k per line or not a full 24 channels, you are getting ripped off. If they won't tell you up front that you have exactly 24 channels coming in, they're cheating you. Also they might not be have a full T1 mapped to you from their T3 (DS3).
Stop the Slashdot Effect! Don't read the articles!
who in the hell lets these gibberish questions onto the front page? no really, i'm a network engineer, that question was incoherent.
The best way to do it is to plug the network connection into a single Linux/BSD machine, and run IPTraf to monitor the traffic (you can grab the URL from FreshMeat.) Then start up MULTIPLE (5+) transfers using things like wget, curl, ab and ncftpget to generate traffic. You should get a sustained transfer rate between 150KB/s-190KB/s (bytes, not bits) Just remember to be a good netizen and don't try to do all those downloads from one site.
Shouldn't be too hard. Walk up to your PBX, open it up, you should see lots of cards. A few on the lower shelf might be labeled something like TMDN64, TIE, you get the picture. Pull them out one by one to see which one might have the span in question.
It is common to see 100:1 ratios in broadband markets. QAM 64 modulated cable modem segments support 26Mbps bandwidth. You will see up to 1000 customers on 1 segment -- i.e. if all customers were maxing out their connections at the same time, they would see 26kbps. And that's just downstream...
The best way to see if your T1 connection is indeed a 1.54Mbps connection is to make it available to the Open Source developer community. Once they have access they can use to host files such a GNU license revisions, pictures of Natalie Portman, and a ton of OGG files. You can then review your Gopher and FTP logs and get a very good approximation of the bandwith used and arrive to the total bandwidth available.
Only by granting access to resources such as T1 connectivity can the Open Source community reach new heights.
Wearing pants should always be optional.
I think the discussion of the local loops is not as important as the data connection type.
:) They ban you to prevent you from discovering the bait and switch the sales person performed on you.
We are in the Internet web hosting and T1 business, and we regularly 'take over' T1's clients from the troubled companies such as WorldCom or Qwest. We have learned a very interesting trick these companies will play on their customers. VERY OFTEN these T1's are not "PPP" T1's but "Frame Relay" T1s. These clients had been lead to believe they had ordered Full T1s, but they were given Frame Relay connections instead.
Let me explain the difference and why it can make a big difference in the performance you get.
A "PPP" T1, is a true dedicated data T1. The data link has a direct connection between two end points. There is no 'SWITCHING' in between your location, and the Host ISP. Using this mode, you will get a full 1.544 Mb/s between yourself and the Internet, assuming they are not over subscribed as someone pointed out already.
A "PPP" is higher performance because it is dedicated line for just your company and it is therefore more expensive.
A "Frame Relay" T1, uses shared data links between the two endpoints. The T1 loop
is carried to a Frame Relay SWITCHING fabric. From there, your data will go through many switches between your location and the Internet where it will eventually get to the Internet. The basic point is your data link is shared and there will experience packet loss and latency on the trip from your office to the hosting company.
Frame Relay is a LOT cheaper because the lines are NOT dedicated and you share the 'ride'.
The latency difference can be vast. Our PPP T1s, have pings times of 1 to 3 ms. On Frame Relay connections you can expect to see 50 to 250 ms ping times.
(This stuff matters if you play quick reflect games such as Counter Strike.)
It also matters to business who attempt to do VoIP or streaming or other
business applications.
It also matters when you get packet loss. With all the extra HOPS along the way through all the switches, you can sometimes get severe packet loss.
So these customers had order 'full T1's but got a shared cheaper frame relay connection instead.
The most deceiving act these large companies do is setup your CSU/DSU (your digital
modem) with the 24 channels you would normally get with a real PPP T1. They
then ban you from having access to your router, so you can not see how many channels have been configured on your side. It is also trivially easy to discover that you are running the Frame Relay protocol and not PPP.
Even worse, is that some of the clients permanent virtual circuit (PVC) is even more restricted. We have had some customers, get a 24 channel CSU/DSU setup (thinking they got a full T1) and we discover their Frame Relay circuit is only giving them 384 K or slower connection. They usually get very angry at this point.
So, after we have made the switch from their Frame Relay circuit to a non-oversubscribed PPP T1. The performance difference is night and day. The customers can not believe how much FASTER everything just became.
I don't know if this site is dead on accurate, but I know that it seems to be better than any site I've tried before. Go to bandwidthplace.com/speedtest Try it at night. The server is in Texas and I am on the East Coast and it still seems accurate.
SIGFAULT
ftp://ftp.src.uchicago.edu/pub/OpenBSD/3.1/src.tar .gz
:)
The math is simple. 1540 Kbit / 8 = 192.5 KByte.
If the line is empty, and you dont get ~ 190-200K, you dont have a full T1
Step 2: Download, late at night, 10 parts of a very large file simultaneously
Step 3: Add up the sizes, divide by the time it takes to complete
Or more specifically.. download all 10 parts of Visual Studio 6 SP2 at 2am PST while you are blitzed on caffeine. Add the sizes (100 MB) divide by the time (3.5 minutes! Double check your math multiple times. Celebrate your cable modem speed with large amounts of liquor (no this is not optional!)
For the mathematically challenged:
Now this gives you effective download bandwidth which will be less than theoretical bandwidth due to network overhead but for a T1 you should be getting in the 1.35-1.4 Mbps range (175KBps or so). The upload speeds can be tested a similar way but you need someone with better bandwidth (say the other companies that are trying to sell you a T1) to download a 120MB file in multiple pieces off of you. Of course they are going to lie to you about the speed the detect so this wont work but you get the idea. My cable modem at home at night (not the same one described above) will typically max out any T1 at 175 +-5 KBps, so this is a valid testing method as well (hey just send me a consultants fee and a URL to test and I will do it for yaThe other part of the problem is that you are sharing the T1 with 2 other people through a hub. Unless you have a way of preventing them from using their bandwidth for the 20 minutes or so when the test would be performed, you will not get valid numbers. There are a few ways to go about this, from convincing them that you need to see if the company selling you the T1 is cheating you all, to "accidentally" unplugging their network cables.
In any case, I wouldnt trust DSLReports to give you valid numbers, but its a good first approximation plus or minus 50%.
There are many alternatives of course, but use the evil empires resources for good! Test your download speed at microsoft.com today!
P.S. Getting the math to show up somewhat legibly was such a pain.. why cant we just use the PRE html tag? *sigh*
I have to agree that any ISP/Telco that tells you they *don't* oversubscribe is full of sh*t.
More than likely, if you are even getting 900Mbps, you are getting a T1's worth of bandwidth from the ISP to your building. That certainly doesn't guarentee you a full T1 of bandwidth out of the ISP though.
As has been said, a SLA guarenteeing that bandwidth would be a legal way of guarenteeing the bandwidth (a contract... if they can't guarentee it, they won't want to give you the SLA).
The possible advantage to a way-too-big Telco is that the big names generally have more customers and therefore more money coming in, and thus need bigger pipes. I know at work we had proposals in for our corporate network from several large vendors, one of which gave a good spiel on their backbone of OC192's across the country. Now, if you can get a T1 into an OC192, they'd have to pretty much be loaded with people going full-out bandwidth to peak that baby. Chances are, even if they are oversubscribed, you could get a full T1 90% of the time.
Interestingly, this same vendor made a point of telling us that they recommend we take our Frame lines and don't do a full T1's CIR. CIR is the *guarenteed* bandwidth (which likely is what they use to gague the OC192 subscription)... we have several sites that are at a CIR of 512 or 768, yet I've seen them peak well above that at times. You can still get *up to* a full 1.5Mbit T1/Frame speed.. its just not guarenteed. And getting a CIR of 768 cuts several hundred a month on the price (and we infrequently use more for that site). Just be sure you are getting a CIR, and not a rate *cap*, which will limit you to that rate.
The supposed rule of thumb for upgrading networks is 40% utilization according to equipment vendors like cisco but of course ISP's tend to wait till it gets to 80% for obvious reasons of cost and since 40% is clearly a way for vendors to sell more equipment. 80% is probably pushing the envelope a little but if your stuck in comcast hell like me on a cable line you are used to a provider that refuses to upgrade even at 120%.
First of all you may see improvements by using a switch instead of a hub (of course its probably that he does not know the difference and is using a switch already) If so you should definitely be using a switch that can divide the ports into VLAN's or use seperate routers so that the 3 office networks can be seperated for security reasons.
It seems your concern is that your ISP has a congested network that is over capacity. If this was your own line and not being shared the way to find out would be to see if you can get full t 1 speeds during busy times of the day but since it is being shared you can only get an accurate test when no one is using your connection.
To be specific, if you want to know if you have a T1 line, as others have pointed out, get it in writing from your ISP. If you want to know what your total throughput is on average, do you want to know the "real world condition" throughput, or the to-backbone throughput? If you just want to make sure you're getting fairly close to normal bandwidth(and not avery bit you can just cause), check out a web-based bandwidth meter. It's not super accurate, but it's not bad. Check out www.toast.net. They've got multiple test sites, so you can get an average based on proximity to your location, etc. If you really want to maximize the accuracy of your results, you'll want to be the only computer on the line, and check your MTU, TTL, etc. settings, just for the hell of it(though it might be overkill for a test that's this loose to begin with).
--Not to be worried, Pitr fix.
If you're going to transfer files to see what you're "true" speed is then you should use TFTP instead of FTP.
FTP uses TCP acknowledgements (to make sure packets are sent reliably) which eat up bandwidth. TFTP uses UPD which doesn't check to see if packets were lost. TFTP isn't 100% accurate, but it is much closer than FTP.
"A plan fiendishly clever in its intricacies"- Homer Simpson
sh int s0
This will show you info like your line quality, errors and the rate of the bits going accrss it. 9 out of 10 times your T1 will operate full capacity, the big question is what kind of network and peering does your provider has. Are they over subscribed?
IE. My cable modem can do 27 mbps, but RoadRunner doesn't have the capacity to handle that.
The peice of cooper that terminates at your site will almost always be the same wire that all the T-1 providers will provide. It's just depends which cage on the other side of the wire you hooked into.
zin
-ZiN-
Disconnect the other two tenants and plug straight into the T1 (I'd recommend at an off hour unless you ask them first)... see what you get.
Damn.. see, its just you and you *alone* wanting the whole T1.. and then you let the other tenants go in on it and you aren't getting the whole thing. WAAAAH. Geez... even *YOU* "oversold" the line by getting them involved.
Nope. Switches were designed to improve bandwidth usage, not security. It is trivial to get a switch to forward frames out a port it should not.
Yes most of the bandwidth testers I have seen have real problems with things over about 700 Kbaud, and their accuracy is a problem at the higher speeds. I have found that FTP to a site you know is fast works well, you might try NCSA. And you want a large file (couple of hundred Meg works best, need to average over a number of minutes). Also contrary to what a person said previous it is not 8 bit/byte.. there is some overhead with IP and possibly transport, I use 10 bit/byte which is a bit high, but I can divide by 10 easy.. (most nights) You might want to do this in a quite time (say 4 am is good), but anytime of day should work.
IF you are a real fussy person you can do the following: take ping and ping say 20 times with a packet length of 64 bytes - take the fastest number, repeat with length of 128, 256, 512, 1024 bytes,again with the fastest times. Get your self a page of graph paper (okay, I suppose for slashdot people you can use your hp graphing calculator) Plot bytes going up, and time going out, it should be a straight line within the accuracy of your pinger (I would recommend using a linux system for this, windows is a bit course). The slope is speed (bytes/sec) without overheads, if you note, it doesn't go though (0 sec,0 byte) - the offset (0 byte, time value) is going to be your latency. You can do this with any node you can ping (so you can go past your ISP to test your ISP's upsteam ISP), remember this reflects the slowest link.
It is possible to work backwards and find all the speeds (and latencies) for links from you out (with increasing errors).. but that is more complicated and for another time...
Isn't the more important question: Is the bandwidth meeting your needs? Are you getting it for a good or at least fair price? If it's meeting your needs and you can't do better elsewhere, does it really matter if you're getting every bit of bandwidth that's possible?
If you're starting to strain the limits, then a simple bandwidth meter should be able to tell you how much is being used, at which point it would be pretty obvious if you're getting what you're told.
It's hard to say for sure because there are too many variables to calculate. What you can do is measure over a period of time and then take a best guess at whether or not you're getting reasonable throughput for the price.
But that's just MHO.
My advice, having done it before.
There is no such thing as overselling. He is selling you a T1 to his network. You will get precisely that. Everything else, you can tlak all you want about whether it's "Good" or not... or oversold or not... but when it comes down to it, you need something you can put on paper, in the contract.
Find out who your ISP is connected to, and how. Get them to make certain latency guarantees to certain points outside their network. None of this is exact science... so decide what the level of service you are willing ot pay for is and then ask for it, come to some agreement. Don't just leave it hanging. I mean, okay, if it's really really cheap, and he says "Well it's good right now" go ahead and gamble.... if you want.
There is simply no easy, clear way to define it. ISPs are not overselling when they get a T3 and then sell 60 T1's... they weren't offering you a T1 to their upstream... if you wanted that, you would just go to their upstream directly, and get a better deal.
ah, was refering to an evil neighbor promiscuously listening.
Why spend lots of money with T1 when SDSL (the Newer T1 service) can do it for you. It *IS* exactly like T1 service without the extra costs of Regulated T1 Line costs. However, if you really want to know the real deal, these big guys have no push anymore as they are also bottlenecked at the border routers. So it makes no difference if you go with the big guys or the cheapest price, you WILL be limited in speed one way or another.
*Headline News* censorship shuts down the Internet! More at 6PM!
There's an open source tool for that. It's in Debian.
apt-get install bing
Why does it matter if you are getting split off of a DS3? Hell, a DS3 is just a buncha DS1s anyway. Even if it's not split off of a DS3, it's gotta be split off of something--some switches can do DS1 drops directly out of the high speed ring.
My friend just got SDSL and it is about the same speed that you guys are talking about (1.5mbps down/765kbps up) I dont know what I am missing but why pay that much for a T1 when you can get higher end DSL for about half? What is it about T1 that everyone likes so much? On another note I get about 2.4 Mbps on my CABLE modem down, and I only pay 40 a month. But my UL is caped at 400kbps. What are the normal speeds of T1 lines up and down? I had no idea T1 was caped (?) at about 200 kBps! That is slow compared to my average 250-300 kBps on a cable modem. PS go to DSLreports.com and look at what school has the fastest conection. Go bears!
"The End of Hell as we know it predicted"
"One billion obligations fall due as Hell freezes over"
"Snowball gets new chance".
There have been several good suggestions about how to test your T1, but you may have noted that some imply you'd be getting 5 T1s and testing them all so you know which one to keep, which is somewhat unrealistic.
Perhaps the obvious overlooked tidbit is that you're sharing this T1 with other users via a hub.... and it needs to be said that you're at far more risk of the other users siphoning off your bandwidth than you are of your ISP not providing it in the first place. To do any of the suggested tests, you'll want to be disconnecting your neighbours first.
I might suggest a bit of cautious thought here. We are one of those small ISPs, and naturally we buy bandwidth from the bigger ones. Recently, one of them tried to FUD us into buying T1s because another of them is financially insecure at the moment. We told them that we'd be happy to talk to them (and we do intend to deal with them), but don't bother trying to sell us by FUDing the competitor's service, just tell us about the merits of their own and end it there, thank you. We can fill in the blanks for ourselves. We tell everyone that, and I recommend you do the same. If the small guy swears he's not oversubscribing, do you have any reason do disbelieve him besides the FUD of a commissioned salesperson trying to meet his quota? Try asking for a reference or two. Besides, as has been pointed out, oversubscription is not inherently evil, as it's based on the fact that not all of the pipe is in use all of the time.... the same as telephone line pools. It can be done properly. If you're measuring your bandwidth in anything less than sustained Mbps, I'd suggest you won't notice if you're only able to utilize 90% of your 1/3 of the T1, as the difference here is probably less than if your neighbours both decided to run "Windows Update" on their PCs at the same time. One thing worth remembering is that people who are actually able to browse the Internet over a really BIG pipe tend to discover that a lot of the sites on the Internet just aren't that fast. If you really do notice, the logical conclusion is that you and your neighbours are oversubscribing the T1 already, so you really couldn't criticize the ISP for following the same practice, if he is.
The academic side to your question has good merit, and I believe has been answered. Honestly, in your situation the practical side is probably moot. If the local guy checks out with references and offers a good product at a good price, deal with him - give the little guy a break while he competes against the incumbent telcos, a position which is not for the faint of heart. At least he'll know your name when you call and you'll probably get through to the sysadmin if you need support, rather than having to deal with "Operator #17" who can't find your ticket number. If the local guy can't deliver, by all means go with the telco.
My $.02.
-brt
Penguinista!
You will be un-assimilated. Resistance is just plain stupid.
BTW, if you plan on asking for an SLA you need to ask for it in writing in this manner to protect your ass. Otherwise, you will never get proper SLA refunds.
- All SLA tests will be conducted from the CPE Point of Entry to the Provider and the Border Router to the CPE equipment.
- The average Latency difference between Provider and Customer will be calculated in this manner (CPE Ping to destination - Provider Ping to destination)
- Should there be an average latency over the SLA agreement, the customer will get a pro rated refund for the duration of the issue until the situation is cleared.
- This SLA agreement is ONLY within the Provider's Network. Sites outside the provider's network will be calculated to the Provider's Border router.
- Should the Return Paths be available (via remote end doing a traceroute back to the CPE) then the Point of Entry into the Provider's network will also be tested. This is to be able to cover the FULL extent of the path.
We have had many issues described above however with Worldcom, they pulled a Homer and said that their SLA server (which averages their whole network to one location with prioritized packets for the SLA server) were not showing any latency issues and therefore will issue credits.
If an ISP is saying to get their SLA, ask them that you would like to have the above clause put it. I will guarantee you 90% of them will say sorry we will not. Therefore, these providers with current SLAs are just making the extra buck on a cheap SLA they have which will NOT be advantageous for your extra spent dollar.
Trust me on this, I have tried to get the higher priced SLA providers to add my clause in and they all refused. SLAs are worthless unless there is another major catastrophe that takes out half the US Internet market.
*Headline News* censorship shuts down the Internet! More at 6PM!
I have to take issue with this whole "banning access from the router is a trick being pulled on you".
... we'll give you everything you need to set it up and even support you on getting yours configured if it's Cisco gear.
The ISP I'm currently working for does not allow customer access to the router unless it is *their* router. In that case, we don't manage it either
We will, however, give the customer an MRTG graph to demonstrate usage on their line.
As far as Frame Relay, the scenarios you describe are the rule rather than the exception, but if you have clear (and private fiber) between where you're mapping ports, the end user won't be experiencing that sort of latency. You're thinking of the ultra ghetto fly by nights that do that.
As a vet of both the ISP and telecommunications industries, I can vouch that every carrier oversubscribes their bandwidth. Trust me, there's nothing wrong with that. There are actually formulas (for both switched voice and data) which will tell you how much bandwidth to is necessary to support n users. You also have to bear in mind that even with a straight pipe to Mae East, you still will never be able to get that amount of bandwidth. A percentage will always be occupied with retransmits, etc.
As far as a real world test, the best thing to do is go to the web site of a university in your city/state on a Friday night at 4 in the morning and download files. The colleges are on state sponsored networks which give them enormous amounts of bandwidth (much of which isn't actually used...justification for inflated budgets) This will allow you to ensure that you're getting the cleanest connection to a high bandwidth remote site.
I actually contacted iStop myself about reselling my (residential) bandwidth, they told me "It's your bandwidth, you can do what you want with it". And that's my _RESIDENTIAL_ connection (3500kbit down 800 kbit up).
So as they said, you can do what you want with the connection. iStop operates in montreal, toronto, and ottawa.
For their website, with price plans and contact info, go to http://www.istop.com/
Regards, Guspaz.
Thanks!
:-)
Too bad you don't operate in the area I plan to open up in (Kitchener/Waterloo)... But at least now I have an idea that there are ISPs out there willing to serve people with needs like mine at a reasonable price.
If you could be told what you can see or read, then it follows that you could be told what to say or think - BoC
your local loop provides a continuous pipe of 1.5MBit/sec to THEM, which has a max theoretical bandwidth of 192KB/sec. However, even w/ a cisco router doing T1 over a cross connected v.35 cable (to itself), you'll never see TCP traffic over about 150KB/sec. It is physically impossible for them to "steal" this. Now, just as certain as you get the full pipe to him, you will never get his full pipe out. If people did not oversell the internet would be mathematically impossible. There simply is not enough througput available on the planet to guarantee no bottlenecks between you and anyone else you would actually perform a transfer with. Look for a large provider that is well cross-connected, like Internap. The more networks a provider has direct access to, and the better the routing, the better your overall performance will be. If you are on a tier-3 provider, you may be blazingly fast to some places, but getting your packet across town may take it though abu dabi.
----- Refactoring is the reason why man does not mistake himself for a god.
Company X says Company Y is overselling the # T1's they're putting on the DS-3... YES they are so is everyone else. The question is it to a sustainable level. Most of the time the T1 your business uses is not full 100% of the time. This means in the down time what you'd use out the DS3 can be used by other businesses. There's always less upstream bandwidth to a provide then what they're doing for downstream. Additionally, you have to consider what if your traffic stays local to the ISP network itself. Sure you pump a T1 full but 50% stays within the ISP network and never leaves the DS3.
Cheapest cost does not always mean best. Use interviews of their customers, look at the SLA's. If your current provider is meeting your business demand is their price really that far off? If you don't own your IP's you'll have to change them all. What a pain.
Comment removed based on user account deletion
if you can download at 150-180 kbps a sec, and also uipload at 150-180 kpbs a second (try connecting to various places to make sure you've reached one where you can reach the maximum), then you have a full T1. Other wise, you don't.
This seems like a pretty easy question to me, maybe they should hire someone competent instead and you can assist them...
You need to first figure out if you are buying a T1 link with Internet access, or are purchasing Internet access as T1 speeds.
It may seem subtle, but there is a big difference. In the first you have a 1.544 Mb/s link (either a true T1, FrameRelay or rate limited connection of some type) to the ISP. They in turn allow you access to their backbone connection. Here your ISP guarantees 1.5MB/s to their POP, but not necessarily to the Internet.
In the second you purchase a link (perhaps a T1) and Internet access at a set minimum bitrate. Your ISP guarantees that you get from your site to the Internet at 1.5MB/s. Once past their backbone router though, they aren't responsible for performance.
Also realise that you will never achieve that throughput for payload data. Each 1500 byte packet has a significant amount of wrapper data for all the lower protocols (IP,TCP,Link).
As far as your ISP's claim that they aren't overselling their T3, I don't believe them. I've never encountered a provider that didn't oversell their backbone. I'm not saying it isn't possible.
For testing: All you can really test is your link speed from you to your ISP. If they have an FTP server, transfer a large file and time the throughput. Once you get to the Internet all bets are off for throughput testing.
A comprehensive program of throughput testing and ping times from servers across the net at many different times of the day would reveal your maximum Internet throughput, but you never can tell exactly what could be causing any slowdown.
If you suspect your link is slower than it should be, ask whoever owns the local loop to run a circuit test, this will usually be the local telco. Tell them you need a local data loop tested for noise compliance because you are getting low throughput. They should do this for free as basic troubleshooting.
Article X: The powers not delegated... by the Constitution...are reserved...to the people
a switch creates an isolated connection between two machines. This does not allow an isolated connection of a bunch of machines to a single router. The advantage of a switch is that if you have a bunch of computers communicating with a bunch of other computers (not to a single router) the switch will isolate any connection between two machines, so other machines can use the network without having to wait for those two machines to send their data.
Your network at school sucked because a T1 isn't enough for most schools. A lot of people assume that a T1 is ultra bling bling OMG cum ur pants fast just because there's a 'T' and a number after it, but most consumer dsl and cable services reach and a lot of times exceed T1 downstream speads.
Joe B from Gobosh?
Either way, you're pretty sharp. It's Derek from Idaho if you are Joe from Portland.
Best of luck to you.
D
I am just amazed at how many uninformed people are around here. And they all think they are experts!! And you people make fun of sales geeks. At least they KNOW what they don't know!!!
Furthermore, SDSL coverage is spotty and rare compared to T-1. Standard DSL distance limitations apply, and the local teleco has to have some fairly new equipment. For businesses (the main consumers of T-1s), SDSL pricing is far higher than for residential. T-1s tend to be more expensive (though not by much), but the level of service with a T-1 is generally much higher than that of a DSL line, which makes for a much more convincing business case.
In short, SDSL is *NOT* "exactly like T1 without the extra costs ".
Put the small vendor under an electron microscope, but believe everything the "big vendors" say. ..and people complain that big corporations run everything.
The Big Guys are overselling. And over and over.
There seems to be some confusion on this issue here. First: the T1 itself is a telco service. It's raw data rate is 1.544Mb, but with you'll never see that as you'll be running PPP or HDLC over it to carry your IP traffic. Minus the protocol overhead you're left w/about 1.4Mb. If you are really concerned about the bandwidth on the T1, you can ask to have access to a local FTP or web server from which to download files and check the bandwidth utilization on your router's interface. When pings climb to the several hundreds of milliseconds, yer maxed.
However, what you're after is probably the bandwidth to the internet. This is a bit trickier because it's hard to isolate just your ISP's connection. Here you can attempt to download a large file from a well connected site (think linux kernel mirrors) and check the bandwidth on your router (so it won't matter who else is on yer T1). Pick a site close to you (traceroute is yer friend) and ping test it. Your ISP's upstream is saturated when pings to this site climb above 100-200 ms.
the only reliable method to verify bandwidth is at your routers interface. you can use MRTG to make pretty graphs if you like. The only way to be certain the bandwidth is maxed is to watch ping times climb far above normal. When a connection's bandwidth is saturated, ping times climb. I'm a network admin at a small ISP. I've done this before, and this is how it works. Just remember that at least on Cisco routers, the bandwidth at the interface is recorded in 5 minute averages, so you'll need to run your test for 10-20mins for any accuracy.
ftp uses tcp, which has a lot of overhead. In reality, a good rule is to divide by 10, and if your getting that, your doing damn good.
Lawyers, MBA's, RIAA? A jedi fears not these things!
The PCI bus has over 50 times that capacity.
That is understating it quite a bit, considering you can run Gig-E on a PCI bus.
ATA-133 is way more than that as well.
http://www.megger.com/us/products/ProductDetailsBy SubGroup.asp?ProductGroupCode=T102&ProductSubGroup Code=T102&ProductGroup=T1%20Testers&BusinessSector =T&BusinessSectorName=Telecommunications%20/%20Dat a%20Communications
This may help: Click here
or is that Here
so much for my +5 funny... forgot the frigg'n closing quote in the parent! I'm doomed to redundantville or flamebait town now.
Are you getting a Full T1. Good question, but first your going to answer a couple of other questions. Like what is a T1?
A T1 is a point to point pipe. From the DS3 to you is a full T1, unless the telecom provider is being dishonest.
1.54 Mbit throughput (probably what you are more interested in) is dependand on several factors. Like do most of you clients use your telecom providers internal back bone? Is you telecom provider a in MAE East and MAE West, or are they buying Internet access from someone else to get on the NET's main backbones?
The status of your providers backbone and the occupancy or congestion at his interconnect points will determine if you get full throughput to the Internet.
If you want to really know what is going on and you are big enough to have some leverage with your service contracts, demand router logs and diagnostic information for his backbone. You are going to have to spend significant time figuring out his internal network structure and interconnections in order to identify the appropriate logs to review. If the provider does not want to produce them, leave and go to someone that will provide you a level of honesty that you are comfortable with.
O'Neil.
I've found in my case that DSLreports.com has usually underreported my bandwidth by 25-50%. I have a bandwidth meter installed on my system, and I would run a bot hammering my ISP's newsgroups for about 150 KBytes/sec, and hammer mp3.com (they always have AMAZING bandwidth) for music files for another 300-350 KBytes/sec by downloading 10 files at a time. This resulted in a throughput of 450-500 KBytes/sec over cable, even though DSLreports had reported around 250-300 KBytes/sec just a little earlier before I started hammering with the other stuff.
did you ask your mom? she told me how to do it. heh.
Install a bandwidth meter on the machine that all the traffic flows through. Remove any software filtering out pr0n sites. Then let a few teenagers into the office for the afternoon to surf. If your line isn't maxed out at the promised service level, then you're getting ripped.
Or just go over to mp3.com and keep a constant flood of 10 mp3s downloading. They have never had any problem pounding bits down at 400+ KBytes/sec, or 187.50, now that my cable is capped.
T1 is a specification to deliver voice or data over copper or fiber, but typically handed to customers as copper with an RJ45 head. (usually the same type of cable as ethernet, but different pinouts)
True T1 is determined by the equipment on the ends of the line, and potentially the equipment owned by telco in the middle. When used for data, it is capable of 1.544 Mbit/second. If you purchase a T1 for internet service, you most likely have a full 1.544 Mbit/second T1 circuit which goes from your site to your ISP's building. Once it reaches your ISP, they connect it into their network somehow. The question of oversold DS3 has to do with your ISP's connection to the internet. Your ISP hopefully has at least one higher speed link to the internet (DS3, which is 28 T1's or about 45Mbit/sec, or higher). This connection is what will limit your actual download/upload rates to the rest of the world. If this link is significantly 'oversold' it is possible that you will not get the 1.5Mbit you are paying for...
The problem may be further down the line, though, and out of your or your ISP's hands...Your ISP will hopefully have relationships with their larger ISP's, but your ISP can only guarantee bandwidth so far. As another poster mentions, the maximum transfer rates you can get are limited by the smallest link along the way...(which may be less than T1 speed)
It is possible that the technology to deliver the service is not in fact T1, but that is not clear from the initial post. If the ISP controls the router at your site that terminates the equipment, they most likely hand off to you as ethernet. If thats the case, they may be using DSL or something on the other side that isn't, technically speaking, T1, but still capable of giving you the same basic data rates and speed as a T1.
Good luck! I think if you are able to regularly get 1.2Mbit/sec total to somewhere out on the internet(using one, or several of the recommendations made as comments) you probably have nothing to worry about...
It all depends what you are after. I would go for the cheaper service if I was satisfied with them, especially since I have a knee-jerk reaction to salespeople who try to shuffle the last fifty years or so of traffic estimation techniques under the carpet to sell a more expensive solution.
Other people have told you to ask for the usage graphs of your present provider. That will tell you if his resources are maxed out for any reasonable amounts of time. If they are, take it up with him. If they aren't, give the salesdroid a kick on the shin from me.
I'm a network engineer with the Japanese ISP division of a large international networking company whose name you would immediately recognize but which I cannot use here.
You have really asked two questions:
1) Is the bandwidth oversold
2) Do I have a true T-1?
Neither of these has anything to do with the other. I will address the first question, well, first.
All ISPs oversell their bandwidth. This is a perfectly acceptable and accepted practice, and hurts no one unless they are peaking their upstreams (an ISP I worked at before was doing that every day and it was ugly). Overselling bandwidth is how they make money. It's how the biggies make money, too.
If the little guy is saying he doesn't, he's either lying or he's runnin deep in the red. If the biggies say they aren't, they are definitely lying. If the biggy sales people say the little ISP does but they don't, they should be shot. No, hanged. No, both.
Generally, you'll get better service from smaller ISPs, where you're not just a number.
As to whether or not you have a true T-1, that falls into two parts: Did the telco provision it as one? If so, you do. If you don't trust them, this can be tested. Second, is it a T-1 but do you think the upstream is throttling you? That can be tested, too. Others have written plenty about how.
My recommendation: do what you should have done before you started this process, and hire a qualified consultant to spend a day with you checking it out and explaining some things.
"We will, however, give the customer an MRTG graph to demonstrate usage on their line."
Wcom and Qwest and the BELL company in our area do NOT allow you to do this and/or do not give you the tools to do this. (They will not share the SNMP passwd.) Kudos to you and your company that you do.
Anyways, the customer always enjoyed finding out what they had gotten and how far apart it was from what they ordered!
Still using a T1? We used those ages ago, upgraded a bit since then.
We tested our T1 by sending it out to destroy a small village. It turned out that it wasn't water proof, and was quickly stopped.
We are up to the T1000 now, just about to send it on it's first test to. The lawyers couldn't do anything about them humans reverse engineering our model T101 CPU, even when threatened with the DMCA. So we hope this will sort them out.
Constantinos Dovrolis has moved to Georgia, and so did the web page mentioned above.
different aprroach
are you happy with your bandwidth?
are you happy with the price you pay for it?
that's it.
Netperf is a good way to test throughput, available for many platforms also - even windows is supported, which is rare thing with good tools. ;-)
Unfortunately, you need a opposing server on the other side of the net segment you want to test, and this is really difficult to come up with in some cases. Passive logging and creating traffic is a good way in case like this. MRTG is a nice way to visualize the results, as people already have mentioned few times.
...Two diffrent worlds. Frame clouds are usually 4 to 1, or 8 to 1. Because the average company that has a t1 nowadays, just has it because the admin has a hard-on for geek factor, or they can't get any other broadband solutions. Frames tend to carry an SLA, you're guarenteed 700k/s **TO THE ISP**, burstable to 1.544Mbps. That's the confusion i'm getting. When you get somthing like a cablemodem, T1, DS3, etc. etc. You're only guarenteed those speeds to the ISP, beyond that, you'll have to bitch to said local guy's upstreams. Honestly, the company i used to work for had over 200 T1 customers, i think maybe 10 used more than 300k/s at any given time. The receptionest looking up her family history, or checking email doesn't utilize a full pipe. Now if you're streaming some Video on Demand, well, then yea, you'll fill that pipe pretty damn quick. :)
Lesson of the day: P2P/Frame T1's are only guarenteed to the other end. Once you start routing across the global Internet, your speeds, can, will, and most definately be varied.
First off, some basics: Channel = 65,536 (or 64k) bits of data both ways. T1 = 24 channels. DS3 = 28 T1's. OC3 = 3 DS3's OC12 = 12 DS3's OC192 = 192 DS3's Channels are also known as DS0's. T1's are also known as DS1's. DS3's are also known as T3's. NOTE: In Europe T1's are called E1's and have 30 channels (DS3's are called T3's). With this in mind, a full T1 can transmit 1,572,864 bits per second (or divided by 1,024 you get 1,536kbps) and receive 1,572,864 bits per second at the same time. There are three components to a T1 (or a DS3): A local loop connection to you, a local loop connection to your ISP, and the interconnecting circuit. There are costs for all three parts. BTW: T1's were (and still are) originally used for voice communications, and can have 24 phone conversations going on at the same time. Each conversation basically uses 8,192 8-bit digital samples per second to transmit voice traffic (CD's use 44,000 16-bit samples). What the heck is a Frame-Relay T1? Sometimes an ISP will sell you a Frame-Relay T1 instead of a true or full T1 (FYI: A true T1 has four wires, a frame-relay circuit has two.) Frame-Relay T1's are shared T1's - this basically means that they provide a certain amount of bandwidth and anything above that may or may not make it to the destination depending on how busy their network is at that point in time. This is usually set to 50% of the total bandwidth and is called CIR (Committed Information Rate). After this it starts to get confusing... Most applications, bandwidth testing sites, and ISPs will tell you what kind of bandwidth you are getting, or can expect to get, as bits per second (bps) some tell you in Bytes per second (Bps) and still others tell you the results in characters per second (CPS) - make sure you understand the measurement they are using. And if that's not confusing enough, a byte or a character is generally 8-bits, but not always! Sometimes it's 7-bits, 9-bits or even 16-bits. BTW: The networking industry, due to this confusion, uses the term Octet which means exactly 8-bits. So, to figure out the maximum bandwidth on a T1 you take the number of channels (24) times 65,536 and divide the result by eight (to get octets per second) you get 196,608 Octets Per Second. A lot of people will state this as Kilo Bytes per second, thus you often see the 192KBps (Kilo or 1,024 Bytes per second) figure bandied about. Keep in mind, this is the pure data rate and there is overhead to send and receive data (like the IP protocol wrapper, etc.), so by the time you are done, your data rate will be around 188KBps. Now on to the over-subscription deal: ISP's will generally purchase a T1 from one of the Tier-1 or Internet Backbone providers such as UUNET, Sprint, BBNPLANET, Cable and Wireless (was originally MCI's internet backbone), etc. and connect it into their network - these usually cost over $1,000.00 (Though, I can get a Sprint Internet T1 right now for $930.00). They often have Dialup users, Web Servers, Database Server, Email Servers, and so forth interconnected to this network, all of which can consume bandwidth. Then they sell "Internet T1's" to their customers, for a reduced charge - what they are really selling are T1's connected into their network, which then allows access to the backbone providers T1 through their network. They will often sell five to twenty of these "Internet T1's". This can provide them the opportunity to snoop on any traffic that you are sending across their network, they can also block, cache, or perform traffic shaping, which can allow someone else's traffic to go ahead of your traffic, or your ahead of someone else's. This system usually works out well, as most people don't fully utilize all the bandwidth all the time and most ISP's don't do anything fancy to the traffic, but can cause issues if you don't know what you are getting into. Questions you should get answered from your ISP: Who is their Internet provider? Is their Internet provider a Tier-1 or Backbone provider? (if not be careful) What does their network look like? Ask for a network diagram. How much bandwidth do they have? How many dialup users do they have? How many T1 subscribers do they have? Ask to see their bandwidth usage reports. How reliable is their network? Do they have more than one upstream provider? Do they do BGP4 routing with their upstream providers? (what is their AS number?) Get a network diagram of their upstream providers. Do they own their own IP addresses or have they been assigned them by their provider? How many IP addresses can they give you? Do they provide DNS servers? Can they setup reverse DNS zones for IP address they assign to you? Do you do network caching? Filtering? Traffic Shaping? To sum this all up: You can send and receive approximately 188k Bytes per second (or 1,504k bits per second) across a T1. Nothing is simple. The telecom (and networking) industries have many names for the same thing - which can be used to confuse and complicate simple things.
Ok..here's a brief lesson in the economics of the Internet. Sorry about the AC posting, but I'm too lazy to log in right now.
Say you're a small ISP, or you want to be one. You buy DS1 Internet service from XYZ Telephone at a rate of $X/mo. You want to be 100% TC (Technically Correct), so you resell your service at a 1:1 subscription ratio.
In order to stay in business, you have to make a profit, so you charge $X + profit/mo for your service.
Explain why any sane customer would buy service from you rather than from XYZ Telephone.
By extension, this explains why the Internet sucks so badly--all the small time providers who tried to limit oversubscription to a low ratio, thus providing excellent bandwidth, lost out to the less scrupulous ISPs who went the high-ratio route.
High oversubcription ratio = more profit anb/or lower prices, and since most people wouldn't ever notice/be able to prove the difference and switching ISP's is a Royal PITA, the equation favors the high-ratio providers.
Get your friendly neighborhood hacker to DDOS you.
That will make sure the line is used to its full capacity
As of Postgres v6.2, time travel is no longer supported.
What the competitors are telling you is that this particular provider has a DS3 circuit that has basically 45 Mb/s of throughput. In other words it can support 29.22 T1 users running at full throttle. In reality if you are using your full bandwidth then you are not!!! VERY IMPORTANT HERE!!! If your T1 is loaded then the routers on each end are dropping packets since they don't have very big buffers. This causes packets to be retransmitted, messages to be sent requesting TCP connections throttle down, timeouts, etc. Remeber that the real goal is how many WWW pages, email messages, FTP files, etc. that you can move. It is NOT how many bits per second you can send! If you are running at about 90% capacity then you can consider the line MAXED out. The same thing goes for your provider on his DS3 line.
If you throw in the burstyness of TCP/IP traffic your traffic really maxes out bouncing around somewhere between 65% and 90% of its max rated load. When you add in the fact that people like to have headroom and the size steps between T1 and T3 or partial T3 nobody uses their max bandwidth all of the time. This is something that your provider uses to his advantage by selling more than 29.22 T1's. In reality that provider may have 1000 dial up customers, many more DSL customers, people with dedicated 56K connections, and maybe even some old ISDN connections on top of the other T1's that he has sold. It is fairly safe for a provider to oversell a connection somewhere between 5x and 20x. Especially if you consider that people that have DSL's usually leave them on but don't use them for that many hours in the day. Headroom is defined as being prepared for the slashdot effect!
You need to figure out if your provider speciallizes in retail (home) customers or business customers. If it is home customers then his load will spike in the evening when poeple get home from work and on the weekend. If it is business customers then his load will spike during business hours. Also consider if your provider is hosting very much traffic beyond home users personal WWW pages. If he is then that is bandwidth that is not available to resell.
The only real way to tell is to look at your provider's router logs. Don't just look at averages for a day! Look at averages generated at least every hour over the course of at least a week so you can see when his network (which you will be a part of) loads up. If he consistantly stays below 80% of his upstream bandwidth and will upgrade his upstream connection if it passes that then you are fairly safe. On the other hand if he routinely pushes his max then that is exactly what his competitors are warning you about! If he loads up then everyone downstream from him will slow down as that will become the bottleneck.
Another thing to consider is that it sounds like he has a single DS3 upstream connection. Ask him if that is true and if he has any plans to become multi-homed. This has two major advantages over a single fat pipe: redundancy and load balancing. If he connects to two or more of the backbone providers then the traffic can most likely pick the shortest route to its destination while still having the other one available if one of the DS3 lines goes down. I'm sure the people @ /. have more than one connection if they are served from their business. It is more likely that they are hosted by a large hosting company that is already multi-homed though. A final thing to ask is if this provider does any traffic shaping. This can throttle users that are consuming more than their fair share. But if you are paying for a T1 you should get all of it. Sometimes though a customer will have a partial T1 (this requires a full T1 connection from their site to the provider though) and get to use any excess if it is available.
This is what I used to do for a living so I know a little bit about what I'm talking about! Once again his logs can tell you if he has oversold his service as his competitors suggest. But put some value on a provider that has more than one connection to the upstream Internet regardless of how large their single pipe is. If you want to have some fun ask him how much an OC196 would cost. 8-)
Regards, Tres
Restore America: Dr. Ron Paul for President!
it is how fast the connection is from who you are downloading from or in otherwords your fastest download will be the speed of their upload so make sure if you are doing any serving that they don't cap your upload side to save on their bandwidth cost ~moosie~
How? Have the ISP initiate a ping of death from one of his core routers - Then have someone look at the stats of the T1 on the edge switch => U should see close to 90% utilisation. If the ISP refuses to do so, that means that they do not have enough balls to jam your pipe.
N.B: Posting anonymous as I work for such an ISP (that has enough confidence in its network)
try pipechar http://www-didc.lbl.gov/pipechar/
... why do you really care? If you're getting the service you need, why bother? If you're satisfied with the current bandwidth and ping times through your T1, why buy something more expensive even if it is faster/less congested/has blinkity blinky florbs?
All reasonably priced providers will sell you shared capacity. And overselling is (usually) a question of what the other customers are doing and at what point your upwards link is going to get upgraded. You'll never get full bandwidth connectivity to the providers peering point for yourself unless you're willing to pay through the nose.
But do you really need that or are you satisfied? If you're satisfied, spending more is a waste of money, wether or not your current provider is using IP over avian carrier to connect you.
Try to open ftp and make the ftp public and put some nice stuff on it or make a donkey node or got to irc and share there some stuff i think ppl will test your speed very well....
World Com screws you not by using Frame Relay but by giving you ATM over IP. In other words your new T1 has a capacity of 1.36 Mb/s after the ATM overhead. I know this because my ass still hurts from getting screwed by them! You know what? They didn't even kiss me first!
In reality WorldDom gives you a true T1 into their National Frame Cloud. This part works as advertised. You are correct when you state that it is their frame cloud that will get you in the end - although I have to give them credit - they did not deceive me when I set up my business. They told me all of this. I just didn't know how to route around it at the time. They throttle their bandwidth with ATM which consumes the extra bandwidth. Perhaps my salesperson was more honest than some of the others or perhaps I knew enough lingo that they didn't bait me. I just didn't know enough about my choices.
Restore America: Dr. Ron Paul for President!
It's not bad as bandwidth meters go.
The day Microsoft creates a product that doesn't suck, it will be known as the Microsoft Vaccuum Cleaner!
it:s a ratio 9:1 ... 20:1 whatever.
I would be more likely to trust s smaller ISP to have lower ratios.
I use todo internet provisioning at MCI. It's scary what a marketing run company can shovel off on it's customers. I swear some of their access points were oversubscribed way into the 50:1 range. The only way you'd get true speed of those was to measure early Christmas morning.
-- www.globaltics.net
Political discussion for a new world
Hi
/.'ed and see if you get the same amount
:>
just get a webserver on it get the site
of traffic though the link as the speed of
the link
should be fine so long as the webserver does not crash
I'm surprised no one has mentioned this. Put up a web site, get it mentioned on slashdot and monitor the load on the server.
You know, I think I forgot a zero somewhere :) I believe I meant to say over 500 times, which works out around 93Mbytes/sec.
I've had enough abrasive sigs. Kittens are cute and fuzzy.
Just make him live up to his promise. Make him agree to an agressive SLA. and if you are not getting your full t1 bandwidth, then that should be an infraction of the sla somewhere--latency, packetloss, etc...
If he doesn't deliver, you don't pay.
You have to have a linux machine and a root account to use this method.
root@localhost:~# ping -f www.microsoft.com
Maybe there's some of you that can read japanese and give the rest of us some update, but YahooJP and many other japan telecoms are offering 12mbit DSL service. See: Yahoo!BB Japan
Wouldn't this mean it's waaaaay faster than a standard T1? Why don't they offer this kind of consumer service stateside?
I've done T1 maintainence and provisioning for a VERY LARGE telcom.
From your question it appears you are probably not in a position to know.
Or at least not in one where you can test for yourself.
If you are behind a 'hub' as you say you are then you do not have access to your CSU/DSU, or your smartjack.
You leave more questions than you ask.
My first question is if you know your Access Provider Circuit ID. The second is as others have asked, who are you getting the service from?
Have you talked to your vendor?
Is your vendor also your provider?
In the end _if_ you cannot access the equip. yourself directly, all you can do is trust your vendor and pay them to test.
If your into pain, and simply unable to trust anyone, check out the folks at www.gl.com. The have some nifty PC based T1 testing systems.
Visual Networks (a former employer of mine) makes a product that works like a T1 CSU but constantly monitors the utilization of the T1 link. It lets you see the actual performance under real conditions over a long period of time. Its perfect for what you're trying to do if you can afford their system.
Maybe I did'nt make this clear: I sell telecom services. It's "Certified" Internet Rate on all the contracts I pass from companies like Sprint, Qwest, and XO.
If DSL (or cable is available), there is no reason not to take advantage of it.
While it's true that the 2.3mbit SDSL is not available where you are, iStop offers other services in your area.
According to their website, they offer 3500/800 DSL service (Albeit 800kbit, or 640kbit that you get after PPPoE overhead, is no 2.3mbit, but 80KB/s upstream is still nothing to laugh at) throughout ontario. It's 85$ a month. They also offer the 1.7mbit/384kbit service for 45$/mth.
And if you can get away with it (I was under the impression that if you had a residential line you could get a residential connection. Email iStop to ask, they'll help you out), maybe you can get a residential line. They offer residential 3500/800 service for 50$, and 1200/160 for 30-35$ (Depending on if you get the "self serve" package).
Regards, Guspaz.
PS: I'm not an iStop employee, but a customer.
>How can I tell for certain that I'm getting a full T1? What would an oversold T1 read out to be as compared to a true T1?"
;-)
The question should not be stated as above, but as "when? and to where?". You most certainly have a full T1 between you and your provider, so that means that if you place a machine at the providers location, and you shunt data back and forth between those machines, they will go at full T1 speeds.
What is oversubscription? If a provider has a DS3 (which is 28xT1) worth of bandwidth to the 'net, and sells 29 T1's, then TECHNICALLY he has oversold his bandwidth. Does this scenario REALLY kick in and mess anyone up (ie 29 in a 28 pipe)? Probably not, because the only time that anyone would notice someting is if ALL 29 customers are at a given moment ALL trying to suck in data at a T1 bit-rate, and then, they would find that they would have aproximately 1 in every 29 packets dropped (an aprox 3% packet loss rate). Statistically, to have ALL of those 29 customers that have T1 links ALL sucking in at MAX at the same time just won't happen. In reality, each customer will always think that he's getting a full T1 from his upstream provider. So, when does oversubscription kick in? The ISP should monitor his traffic on his lines (I'm not talking about the lines to his customers, which should also be monitored, but just for service to each customer), and he will notice usage patterns. With enough customers, the average usage on each line will follow a nice series of peaks and troughs. Peaks probably during the day (either business hours, or evening hours depending if his customers are mainly residential or business), and troughs generally during the night. Now, the peaks will NOT be equal in bandwidth to the sum of bandwidths of the customers, but equal to the sum of USAGE of the customers. So, if he has 10 customers on T1's, they will NOT be using 15Mb of bandwidth (how much will in reality depend on his target customers). The provider will see that as he adds customers, the peaks will become more regular (a question of numbers, the more users you have, the more regular the averages), and obviously will start getting higher. If he just carries on selling, and doesn't do anything else, at a given time, the "peakest" will hit the available bandwidth he has. At that exact moment, he will be overselling beyond his capacity. If he continues to add customers, then when he hits peak, it can't go higher (as there is no more available bandwidth), but it will flatten out at the max of his incoming links. All EXTRA people trying to suck in data, will at that point be fighting for whatever is there. If he had 100 customers when he hit peak (and at that moment was EXACTLY maxed, but with no adverse effects), then when he doubles his customers (goes to 200 customers), they will all have a transfer experience of around 1/2 the speed they had when they were only 100, and get worse as more customers are added.
Of course, what people want to know, is "will I be screwed by signing up with this provider?". And that is hard to answer just point blank. A provider who sells T1's to ISPs who are in turn trying to max out their links by reselling to customers, will probably find that each T1 of his customers will be near saturation at peak times, so by having 10 customers, he will probably be needing to have just under 15Mb of available incoming bandwidth. Another provider who sells T1's to small/medium businesses, and sells 100 T1's, may find himself with peaks of aound 10Mbs instead (business users won't be bandwidth hogs, unles there are geeks working and sucking down MP3's, p0rn, software, etc...).
Also note that a provider typically will not have just ONE link, but a multitude of them, which if he knows his beans) he will try to load balance so as not to get any bottlenecks. If however he is maxing out one (or more) of those links and doesn't do something to solve it (increase that link, or move traffic away), then every time your data goes through that link, you will be in a bad position.
So what does it boil down to then?
-note down CLEARLY what you want to do with your bandwidth (just surf? download stuff now and then? Have a quality link to a distant point so that you can send VoIP between here and there?)
-do tests and try to pinpoint where the bottlenecks are
-pray...
My bag of cents...
I see people are posting their favorite network monitoring tools so I might add another one: NISCA. THe advantage of this one is that it keeps all the stats in the db and generates graphs and stats on the fly (PHP/MySQL). You can look at last year's stats without any precision loss (unlike MRTG).
Cheers
However if you are interested in a T1 speed every once in a while (say when you need to download the latest OS patches, stream media, etc..) and the rest of the time is casual web surfing and email then get something from a local provider.
Going with the big boys you will most likley get directly connected to a larger pipe (T3) and will probably not run out of BW when you need it. But you will PAY for that privledge. This is no garuntee of service, just that you will most likley get more BW when needed. I have a major provider, UUnet, and it is a Frame T1, it bounces sporadicaly. I also have a local provider that has a T3, their like is rock solid compared to UUnet. So it is a crap shoot as to what you will get.
Most small buisness do not need that much BW. When I sell BW, it is metered and the clients know that they can burst up to a T1, but it is unrealistic of them the expect 24/7 T1 speeds at a discounted rate.
~Sean
What's a user to choose? Lots of good tools have been mentioned, but to any of them realy fit?
MTRG assumes you have routers that speak snmp, but what if you're using non snmp hardware. (Like a Linux box doing true NAT) The site and FAQ's aren't very helpful in that case.
Cricket looks good, but have you tried to install the thing?
Pathrate measures capacity, which _might_ be useful, but only if you control a server on the other side of the link.
What's a small network administrator to do? What can you run on a 'nix box to track network throughput?
Thanks in advance.
um, he also said he's splitting it out between two other companies and sharing the bandwidth.
I don't know of any effective way to test this, because packets sent over TCP are a very unreliable meter, unless you have a connection at both ends and are only sending network traffic between two endpoints. Rerouting, packet loss, packet queuing delays, bad routing tables, slow machines or network connectors, dirty lines, and any number of other factors could skew the results.
It's possible to get a rough estimate of bandwidth, nonetheless, but it's not easy if you're splitting the line. If you know no other computer is online and running through the hub you should get close to peak performance, but it's still tough to say if you're getting shortchanged because of external factors.
Run pathchar for several hours, take samples at different times of day. It's pretty damn accurate!
s /pathch ar/
:)
Pathchar here:
http://www.caida.org/tools/utilities/other
My solution was to get a colo rack at Verio, and install a point to point t1 circuit from there to our office. I know that I'm not oversubscribing myself
I want to go into business with Billy the Janitor! He has backbone, he has drive, and crack money to finance the op! I could help him cheat on the technical issues with some caching, layer 4 routing and a $300 month satellite uplink, but the man has the right idea! Go Billy!
Think about a brand new house. When you lay out how many amps of electrical service to 'buy' from the electric utility and in terms of the power panel that you install, you don't ask 'what if every outlet in the house was maxed out?' but rather, what is a realistic maximum utilization for the size of the house and the major electrical appliances? The same thinking goes into designing your neighborhood substation and out to your multi-state and international chunks of the grid. Whereas everyone flushing the toilet during the superbowl/world cup halftime won't 'take down' the water system, everyone in a state turing on every electrial appliance in their houses (and firing up all the industrial uses, too) could take down the electrical grid. Yet, with some crafty engineering (peaker plants, pump storage) the system still works with this supply/demand model.
What are the plusses and minuses of the analogy? On the down side, network traffic tends to be less predictable and more 'spikey' than electric demand. On the plus side, too much network use just bogs things down, but too much electrical demand blows the breakers (at various scales).
So does it really matter if an ISP 'realistically' 'oversells' their pipes? Your house/business is on an 'oversold' electrical network, and for most of us, it works pretty damn well. Perhaps it comes down to a realistic expectation of degree of performance.
...even though many of the other answers are correct in their own ways. This is the one that counts.
Say you live in an affluent neighborhood on the shores of Lake Washington in Seattle. Many of your neighbors are Microsofties who compete with each other on who can say their home has the highest bandwidth (hell, they look down their noses at any coworker who has only 220-volt electric service). They don't use all that bandwidth on a continuous basis. But they get great download times as long as the other end is fast. Your provider serves all these sites and has only enough backbone access to handle 50% of the bandwidth he promises if they all use it at once.
Now imagine your brother lives in a seedier section of town where all his neighbors are running pr0n sites in their spare bedrooms. He uses a telco provider which has the system wired to handle 90% of what they promise. But his neighbors use near 100% of their T1s 18 hours a day.
The result: The massively oversold ISP gives you better bandwidth than the not-oversold telco gives your brother.
You can test for this in any number of ways (lots of suggestions have been posted). But you still don't know the answer to the really important question: Am I getting what I'm paying for?
The reason: Suppose the kids of the MS employees one day discover P2P file-swapping, how to replace their overpriced OSes with Linux, and how they can make MP3s of every CD they own, set up an Icebox server, and listen to their music anywhere, anytime. All of a sudden, they are using all of that bandwidth which was just a status symbol for their parents.
The real question for an ISP is how fast do they realize this is happening and add capacity to their backbone connection to prevent bottlenecks. The only way to know is to watch your actual throughput over time and see how it holds up. It's easier for a telco to do it, sure. But whether they do or not is unclear until you have lived with them for a while.
I have a very long story about how my small ISP was bought out by a backbone provider who then bought out a telco. In the end, I practically had to maintain their router network myself by hand.
You can ask all your prospective providers how they monitor and how often and how quickly they respond to bottlenecks (do so, they may give you better service if they know you care). But ultimately careful monitoring of your throughput (particularly as you come to use your whole T1) is the only way to know for sure.
I buy Internet access for my small business the same way I buy phone service (local and long distance). Once a year, I ask for bids (including from my current provider). The telcos tell me their competitors will give me poorer quality service. I tell them that's been the exact opposite of my experience. If I don't get the service I've been promised, I go elsewhere. You're only bound by a contract if the other guy lives up to his side of the agreement.
If you really care about the degree to which you are getting your money's worth, monitor constantly, ask for bids regularly, and don't be afraid to switch providers. I have been constantly amazed by how much better my service is (and how much less I'm paying) as a direct result of doing this.
By the way, if the telco telling you the small guys don't deliver is Qwest, laugh in their faces.
Eternal vigilance only works if you look in every direction.
Thats kind of flashy there yourself.
Anyone with half a mind knows that consumer and sometimes even business level broadband is just NOT reliable enough for a serious business. One of my companies branch offices used a 1.1 sdsl line and it would be down for AT LEAST 30-45 minutes during the day. For a business that relies on the internet, thats 30-45 minutes of downtime. It got so bad in the branch office they made people take lunch when the modem died. Don't get me wrong it was very fast and a great service when it was working, but its unreliability, btw which we could do nothing about except get a refund for the time lost every month from our ISP, was just crappy.
A few basics first. A full T1 will allways run at full T1 speeds. This can be tested by measuring bandwidth between your end and your provides's end of the T. The concern the more expensive providers bring up is oversubscribing (or overselling), That is, is your PROVIDES's pipe to the backbone fat enough to carry full load from all of his customers pipes to the backbone without throttling.
The only real way to measure that, is to pump data between your end and a large sample of servers that are NOT on his network simultainiously. The sample must be large enough that congestion on their end won't significantly affect your results (since no provider can do much about other people's feeds). 100-1000 would be enough (since the load on any individual end network would be less than a dial-up connection's worth). The hard part here is that you can only be 'pretty sure' unless you could get all of his customers to flood his net at the same time. Normally, the best you can do is repeat the test several times and try to test during your best guess of peak traffic hours.
a more subtle sign of potential trouble is dropped packets. In the above test, a few spefcific targets dropping packets is not a problem (their network is likely the problem). However, if all of the connections are showing dropped packets, and especially if the drops come in 'waves', it's a sign that your feed is being throttled to relieve congestion on your provider's uplink.
There are other considereations as well. One that comes to mind is your provides's router capacity. It doesn't matter if your T1 is up to speed and his uplink can handle it if his router bogs down and introduces excessive per packet latency.
Those effects can be measured by using ping, or (where not filtered) UDP echo packets. Here, the idea is to send many short packets to see if his router can make it's routing decisions fast enough. traceroute can be helpful here as well. You'll be interested in the time your provider's router takes for it's hop.
You're splitting a T1 with a HUB??
Dude, you should be using a router or at least a switch!
That's bullshit. I frequently get 1.4Mbps on my ESF 8BZS BRI. That includes protocol overhead, latency from Uzbekhistan, and the alignement of the moon and stars.
That number would also be incorrect, as the current PCI spec (2.1) calls for 1200Mbits/s or 150MBytes/s.
Pathchar - It sends multiple ICMP Packages of diferent sizes to diferent hosts and it shows you how much bandwidth you get. Don't confuse with throughput...
Install it and give it a try. It's the smartest tool that I have used so far.
You will most always get a full T1 that test ok, from your Smart Jack to the ISP Switch. They often will test this switch internally with you from a junque box on their network to show you that you are getting 1.5mb/s by allowing you to download a large file, that is by far the fastest file to download. Plain text. It will run at a happy 1500 k all day long. Internal network speed, which for the most part in your service level agreement is all you are paying for. The Internet speed you are getting through that T1 is not in any way guaranteed to be at the same speed. They will call the reasons whatever they will, but you will have a tough time getting better service from them. I used to use PSINet a few years ago, and right before I cancelled my service with them, I had a Sprint T in my building and had them come in for a test session between the two to listen to their explanations. The Sprint T Kicked it's ass/ THe PSI Net T would never surge over about 1100K. With a constant transfer speed of about 850K. The Sprint T Surged to 1750K and had a constant download at about 1400K. Same file at the same time on the weekend early in the morning coming from the same site.
PSINet used to be a decent provider. They oversold to the point of being obnoxious. Sprint provided good service and bandwidth. Ultimately I changed over to a lesser priced ISP that does not have as good of bandwidth as Sprint, but is much better than PSINet.
hehe. For the next 5 years anyway.
The man who trades freedom for security does not deserve nor will he ever receive either. - Benjamin Franklin
How do you think those 'channels' are used? You think your TCP session is limited to a single DS0 because your ISP has a T1? That would mean only 64Kbps per TCP session.
I'm not sure what you mean though.. I'm fairly certain that's not what you meant.
Routers use a T1, usually, as a single channel. The fact that's it's channelized is irrelevant, all channels are used equally in round-robin fashion, usually.
The T1 appears as a single point-to-point interface on the router.
Oh, of course, it doesn't HAVE to, and you could be using a channelized T1 as aseveral fractional connections to different locations.. but that's another story.
The insinuation was that TCP cannot make full (or mostly full) use of the bandwidth available to it, and that multiple connections are required in order to get full utilisation. This is not true.
Comic Book guy on the net
"oh hurry up I'm a busy man!"
"Huh, the internet king I wonder if he can provide faster nudity"
"I'm interested in upgrading my 28.8 kilabot internet connection to a 1.5 megabit fibre optic T1 line. Will you be able to provide an IP router that's compatible with my token ring Ethernet LAN configuration"
13 year old white supremacists are shitty web designers.
There is no way that anyone can guarantee you a full T1 of bandwidth to any site on the internet. At most they can guarantee you a full T1 up to their gateway to the internet. Once you leave the ISP's gateway, depending on what sites you are trying to get to and available bandwidth between the ISP gateway and the site, your actual bandwidth will vary. The internet "speed tests" are designed to give you a ball park figure and not your actual internet connection.