Slashdot Mirror
Ask Slashdot: What Is an Acceptable Broadband Latency?
holmedog writes "A simple question with a lot of answers (I hope). I recently had issues with my DSL broadband at home, and after a month of no resolution, I was told 300ms latency (to their test servers) was the acceptable range for Centurylink 10.0Mbps. This got a shocked reaction out of me to say the least. I would think anything over 125ms to be in the unacceptable range. So, I have come to you to ask: What do you consider to be acceptable broadband latency and why?"
I used to work for AT&T Uverse and over 200ms was enough to get a tech onsite to look at the problem.
First pos... Dammit!
Maybe if you're coming from off-continent.
300ms is the typical latency of an analog modem.
Slashdot still doesnâ(TM)t support Unicode after it was added to the HTML standard in 1997.
Over 125ms is definitely unacceptable in the same continent.
...of the broadband wars. All consumers really seem to care about is faster download speeds, so networks offer it - by munging up their network so much that latency is measured in seconds. With the death of the network engineer, people just aren't educated enough to realize that part of the whole broadband experience is getting your packets sent and received fast, not just your GET or retrieve request getting all the data it asked for quickly. If you have to wait more than a second or two for your requests to even get there, then most people are gonna give up and try somewhere else.
What are you using your connection for?
If you're sending emails, then 300 is perfectly fine.
Turn based games would be fine. Real time games would be rough.
Request an escalation of your trouble ticket. No reasonable person would expect 300ms latency as the norm.
We can't really tell you what's "acceptable". That ultimately depends on what you're using it for.
Maybe the right question is, are you getting a worse ratio-vs.-price situation than is found in most markets in your country?
Or are you asking whether or not the provided is in breach of the law because they're offering something so bad that their advertising is deceptive?
I can't see 300ms being acceptable anywhere in North America unless you are on a satellite link, however if you are testing over continents then yes.
Testing to the providers own test servers within the same country seems insane to be that high.
EA David Gardner -"... but the consumers have proven that actually what they want is fun."
If there isn't a perfectly linear tube filled(emptied?) with hard vacuum between their GBIC and my GBIC, providing the lowest possible roundtrip time(that fiber crap can slow your photons by 30-50%), the connection isn't good enough.
I'd prefer 50 or lower, I probably couldn't argue below 100, above 100? Yeah, I'm going to make a fuss.
If there server gives me that as a min, it'll only escalate in other programs, which I'd use that against them.
My 10Mbps cable gets 33/80ms at average/peak. A church I set up with 3Mbps DSL gets 60ms. My old satellite rig got about 500ms (less with modem uplink). Do they keep their test servers local, or does a tracert show a number of hops? 300ms is completely unacceptable for the first hop.
Just where exactly are these 'test servers' in relation to you? What, exactly, was this 'test'? This seems a bit of a worthless test. It's entirely possible your DSL has less than 100 ms latency, but the delay is on the server end or the links in between. This is too vague a scenario to comment on.
My feelings about 'acceptable' latency depend on how much I am paying for it, at what bandwidth, with what level of SLA, and for what purpose.
If you're connecting to the house next door, I would expect 25ms or under. If you're connecting to a tentacle porn henti site in Japan, latency can be upwards of 128ms. In other words, there is nothing magic about broadband that reduces the size of the world or gets around the speed of light limitations.
"Have you ever thought about just turning off the TV, sitting down with your kids, and hitting them?"
For gaming 100ms is shit. For general browsing 300ms is still pretty poor, but not the end of the world on a cheap and nasty connection.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
It really depends on 'to where you measure' and 'under what conditions' and 'what technology'. EG, satellite broadband will just have bad latency, period. Its the nature of the beast. And cellular/wireless can vary all over the place.
But for fixed, land-line connections? I'd say well under 50ms of latency for the last hop, so perhaps 125ms latency max to an in-ISP test server (giving them the benefit of the doubt and assuming 75 ms latency to their test server because its somewhere in the middle of the US).
However, this is 'no traffic' latency: if you are doing a file transfer, BitTorrent, etc, the bad buffering in many networks can make the latency under load much much much worse.
It is also "no WiFi latency": your WiFi connection can introduce all sorts of problems, including bottlenecks etc. So it should be the latency you see when plugged into the wall.
Two other resources I'd recommend you look at: Ookla's Speedtest.net, which is a very good speed tester for latency and bandwidth, and Netalyzr, which is a very comprehensive network tester.
Test your net with Netalyzr
Anything over 60ms screws up VoIP badly. Comcast builds in buffering in the modems to cause latency and jitter.
These ISP's are getting as bad as the crap Dial up guys in the late 90's.
Do not look at laser with remaining good eye.
When you say the word "latency" most tech-savvy folks think about the propagation speed of the technology (e.g. electricity in copper, or light in fiber), and thus assume it's basically proportional to distance.
However, latency comes from other things as well. Serialization delay adds latency, and the lower the symbol speed the more it adds. Multiaccess media adds latency while waiting to transit. Multiplexing anything adds a small amount of latency looking for a time slot.
The biggest culprit? Bufferbloat. This is a term that has been coined to describe the fact that many networking devices have entirely too much buffer. In the best case someone has sized the buffer for the max line rate that device may see (perhaps 25Mbps for your DSL modem, when your link is only 10Mbps), in the worst some misguided engineer thought "more == better" when figuring out how much to buffer, or just didn't care. There are a number of efforts to try and fix this poor situation, http://www.bufferbloat.net/ is the place to start. Basically buffers add latency. A small amount of buffering increases throughput, but beyond that it does nothing but increase latency and generally make the user experience crappy. When the link is full you need to drop packets _quickly_, because that's the signal to TCP to back off. Packet loss is a _good_ thing on a full link.
Try running ICSI's Netalyzr (http://netalyzr.icsi.berkeley.edu/) which will attempt to estimate your uplink and downlink buffering. If you have a "router" in front of your DSL modem it may have some tuning, or "QoS rate shaping" that will help. If it's a device provided by your service provider you may not have access to the settings, and it may simply be configured wrong. With some vendors asking for a different model of device may help, with others, you may be screwed.
The technologies involved should deliver 20ms latencies if properly configured. You should absolutely expect that, but getting them to acknowledge a problem may take latencies over 50ms. If your service provider thinks 300ms is normal, you need to escalate or move to a different provider.
I'd have no problem with 300 most of the time, but I'm not a fan of multiplayer games that require twitch reflexes, nor do I do anything significant on OnLive.
Another place you might have trouble is streamed content -- this is one of the reasons I do not like streamed content. I'd much rather use the model of "download the content, use it, and then discard it", which is much less sensitive to latency, lets you get higher quality regardless of bandwidth (as long as you allow the download time to be longer than the viewing time), and lets you schedule your traffic for times when the network isn't congested.
(Alas, it seems to me that content-owners think of streaming-instead-of-downloading as a form of DRM, and so attempt to make the other options look less attractive. Makes it more like TV/cable, I guess.)
Mars? 300ms latency is getting into satellite territory.
I'm paying for a service, I expect no less than my minimum promised broadband.
Life takes interesting turns, but the most interest is when you're off the beaten path.
This is kind of a simplistic test and I am sure I will be chided for it but if you do a simple traceroute with a free graphical tool like "sam spade" you can get an idea exactly where in the path the latency is. Assuming someone in the next room isn't seeding tons of torrents, assuming you don't have a virus etc etc I think a traceroute would be a next step in troubleshooting this. When you identify the offending hop, you can give the provider the IP address of the bad or overloaded / misconfigured device in the path. Tech support people have no incentive to look at their own equipment and always always always blame the customer first. Sadly they are correct in doing so all too often. Good luck! Mike
My CenturyLink 10Mbps DSL in WA State delivers 65ms ping to a Google DNS server. I get 6-7ms ping to their gateway. It's rock solid unless my connection is saturated. They were significantly oversubscribed and were listing our area as having an "outage" for over a year before they finally got our backbone upgraded, but it's amazing now. You can ask them to switch you from Interleaved mode to Fast mode if your line is decent. That can reduce your ping time significantly. But it sounds like they have some other major issue in your area.
I was lucky, I scored a great contact with the central office tech in my area. I'm able to give him timely notification of outages and things, and he is able to provide excellent service through bad times. Their residential customer service is on par with Comcast.
I keep forgetting my place. Jesus is for losers. Why do I still play to the crowd?
1) Is 300ms actually considered acceptable by some broadband customers?
2) Is 300ms typical for some broadband connections and/or modems?
3) Can 300ms acceptable as a general standard of service for broadband connections?
Personally, I cannot accept 300ms for my broadband, because I need to run RDC over a VPN, which would result in about 0.75 roundtrip for every GUI action I take. A little less than a second for every mouse click? No thank you. So, #3 should definitely be NO.
I consider anything >10ms to servers located within my ISP to be absolutely unacceptable. However I'm on a fibre link so my viewpoint is kinda skewed.
When I was on a DSL link (1998 - 2002) if I got >50ms to servers at my ISP I started looking at what may have been clogging my link (in one case I did a data capture and proved to the ISP that one of their Cisco routers was misconfigured and spewing garbage) and then started planning to lay siege to the ISP.
However given that you're dealing with an ISP that I'm unaware of, are their servers located in the same facility that your DSL connects to or are there other hops that it goes through? What is the layout of their network like? Is there traffic shaping between your ingress point and these test servers?
300ms is only acceptable if you're communicating across one of the trans-oceanic links or you're on Dialup/Satellite (and I think Sat links have improved, haven't they?)
I don't know what test servers you're referring to, but typically when I test a client's internet connection with speedtest.net using the automatically found best server I get results under 50ms. I would imagine anything over 100ms to a nearby server indicates some kind of network mismanagement.
I also have clients using satellite connections. Their latency is typically around 750 to 900 ms.
No sig for you. YOU GET NO SIG!
If you have over 300ms latencies to servers inside your ISP's own network then I would definitely call that unacceptable. With my ISP and fastpath enabled I often get 20ms to servers within the same country. Anything over 60 and I wouldn't be able to feed my Counter Strike addiction ;)
I used to get better than that back in the dialup days using less than 56k-Flex from East Coast US to the Jolt servers in the UK. I would think broadband between you and your ISP's test server would be much less.
I am Bennett Haselton! I am Bennett Haselton!
A good tachyon router and you'll have the bits before they're sent, the way it should have been.
The best part is, even if your ISP doesn't provide the service yet, if you plan to sign up, you already have it!
Personally I'd complain if it was anything over 50 ms for wireline. Wireless you are going to see higher; for satellite 300 ms is probably good (but I don't think you can get 10 mbps over satellite yet). With DSL I had ~13-20 ms reported to nearby test servers (well, ~70 miles as the car drives). Switched to U-verse recently, now I get about 22-30 ms to the same server on a bonded pair (interleaved). Don't think I've ever gone above 30 ms to test servers, but then I've only ever had DSL or a T1 at college (for broadband; I think back in the modem days latency was generally on the order of 200-300 ms, but I don't remember for sure).
The people that made speedtest.net also offer pingtest.net, which gives some decent information on latency.
Less than 30 forget about it.
Generally speaking, on a 10Mbps broadband connection I would expect 1-3 milliseconds to the first hop and a few milliseconds per hop additional inside the regional network.
If you start hopping to other continents or if you're located on an island and have to have a satellite uplink or long haul inter-island or intercontinental fiber then you need to adjust your expectations upwards from there. Figure on a half second delay for a satellite link (accounts for both up and down) - 300 isn't enough for satellite so I think you're not having this problem.
What's your traceroute to various places look like (to their test servers for example, and to other locations around the Internet)? Is your delay right from the first hop or does it come later in the ISP network? (I'll assume the delay is on your own ISP network as the delay is to their own test servers).
blindly antisocialist = antisocial
I just averaged together the data for a few thousand DSL circuits, and it seems that the average response time is in the area of 65 ms. Anything above 150ms is out of the ordinary. There are even a few CenturyLink circuits in there (reseller), and the average response time for those is a little higher, around 70 ms. Usually slow response times are because of an over-utilized circuit, but if that's not an issue here, then you should probably check the signal and margins on your modem or have CenturyLink send a tech to do so.
I do have a 10 Mbps DSL at home with the following ping time statistics:
First hop to ISP over DSL line in Finland: 22 ms
City 500 km away within the same ISP network: 33 ms
International connection to 10 hops and about 2500 km away: 50 ms
International connection over some European countries and over Atlantic to New York (~8000 km): 125 ms
Continuing journey from New York to Tokyo, Japan (lots of kilometers): 300 ms
How far is their test server anyway?
A better test would be to do a trace route to google or yahoo and see how long that takes.
150 and less is decent to a server that's not being hammered, but latency for other things like Warcraft have additional things adding to it not just the connection so 300 is kinda high but not unexpected in some situations.
From
http://www.dslreports.com/faq/694
DSL/Cable 10-20ms
Some points for comparison: https://wondernetwork.com/pings/
With only 282ms you can get a ping from Amsterdam to Hong Kong.
paul reinheimer
Latency depends on your destination. It is limited by the speed of light, and governed by how lousy the link itself is. It's how you sometimes get stories like the 500 mile emailFor some reference points:
A map of expected United States latency from some place in Texas.
Often times your first hop on DSL will be slower... my own network right now shows 40ms to my ISP's gateway. 300ms is my ping time from Maine in the US to Australia.
Another helpful source of references are looking glass servers that will let you drop right into another provider's system and see ping times from their perspective.
SIG: HUP
300ms is usable, but not by much. Like in all things, it depends on what you're doing.
For an ssh connection, that's almost unusable. I'd not want to use it for much of anything.
For an AJAX web app, that will probably be unusable unless it was tested with such high latencies in mind and written by competent programmers.
For gaming, you can forget about it. 300ms is about 50% more than maximum for what was playable for network games, 15 years ago, and it'll probably prevent gaming outright on many modern platforms.
About the only thing that's acceptable for is casual internet browsing, chatting, and email. It'll probably make sites like Facebook unusable, but it's probably Enough for most.
~/ssh slashdot.org ssh: connect to host slashdot.org port 22: too many beers
I am on CenturyLink DSL, 1.5Mbps sadly, and I get about 100ms latency across the country. As good as 90ms from CO to AL. As much as 125 or so to locations further away.
One question you should ask yourself is: What test server is this? Latency is affected by many things so when thinking in terms of how many milliseconds does it take to get a response, you need to have a baseline. When my technition came to install our broadband, they used speedtest.net My latency to my closest test server is 35. If I ever saw 300 here, I'd consider that completely unacceptable for a typical advertised 10Mb wired broadband connection (cable modem, DSL). You can also try testing when you're ethernet cable is plugged in, instead of wireless to get faster speeds and probably a slight change in latency. I noticed this with a 50Mb up and down link from Toronto, the wifi couldn't utilize that efficiently.
Sometimes that isn't possible.
I use CenturyLink out here in the sticks, and my wee 8Mbit line averages 36ms to the nearest test server according to speedtest.net. Picking something known to be slow, such as a former employer's UEN (Utah Education Network) server gives me 77ms. SanFran from here (Oregon) gives me 56ms. Funny thing is, it even feels snappier than my previous Comcast line in Portland, which was an alleged 20Mbit (latencies were dog-slow, 80-100ms at times at level best).
A lot of it depends on the local infrastructure, and your own in-home bits. For example, until I rectified things, my Ubuntu install had some rather horrendous lag. May want to check there too.
Quo usque tandem abutere, Nimbus, patientia nostra?
I had a VPN customer on CenturyLink and a previous network engineer had put their home office LAN on 192.168.1.xxx (which is pretty common). The outlying offices were on 10.x.x.x subnets. One day, suddenly, no one could reach the home office file server. I discovered that there was a whole collection of computers with 192.168.1.xxx addresses on the WAN side of the routers. This, of course, broke the VPN links. He didn't just have them on that subnet but he had addressed one as 192.168.1.1 and up through a numerical sequence. When I finally got through to the chief admin guy (in Portland, OR) and told him he had internal IP addresses on a routable network he responded that the WAN side of our network was his INTERNAL network and he saw nothing wrong with putting a bunch of servers on those IP addresses. Nothing could convince him otherwise, either... because he was studying to take his Cisco Certified Network Administrator test.
We readdressed the home office (that was fun!) and then moved to a better provider; one who at least would listen.
No one ever had to evacuate a city because the solar panels broke!
I live in a sublet house that has free CenturyLink DSL included in the rent (shared over wifi). The service is supposed to be ~768k, their lowest tier. I normally got about 600-700k so the speed was as advertised. However latency was normally fairly high (90-150ms to google) and any problems with the line (like a slightly bad filter) made it skyrocket. In addition any downloads on the connection would make latency shoot through the roof. One 240p youtube stream from one of my housemates would send latency to ~600-900ms. Anything using close to the max of the connection would send it to ~1200ms and start the connection dropping packets like mad. I still have this connection available to me, but opted to pay for my own Comcast Business Class service as I am a teleworker. My advice is go with someone else, quick.
Silence is a state of mime.
... 100ms was perceivable by most everyone.
Of course, latency tolerance depends on the activity: game latency on a fps must necessarily be lower than on an rpg, for instance.
For most latency-critical apps over the internet (media streaming, voip, gaming etc.), 50ms might be mark to aim for -- of course, circumstances might make that unattainable, for latencies tend to accumulate from start to end of connection. Some technologies also imply huge latency -- e.g. satellites -- at least until that spooky action at distance gets widespread use.
Latency is the big dark continent of broadband service: most companies don't even know what we're talking about when the issue comes up... and some just pretend it's a non-issue, or start talking about speed.
I have a fairly good link at home (10Mbps) and still latency gets on my nerves, specially on weekends when everyone is online.
I miss the good days when I used to tweak my 33k6 modem to get _lower_ speeds and lower latency; I wonder if that's possible today with 100Mbps ethernet.
All this is just MHO.
Yea I would think you should be somewhere in the ~100ms range - and that's just barely adequate. To your ISPs server you should be getting under or around 60. I think for online gaming they say you should be under 80 but I could be wrong on that.
For reference, I get 149ms ping on my cellphone with a shitty connection.
or else!
Seems like the marketing goonies got you - 24 Mbps is the theoretical maximum of ADSL2+. Kinda like how the theoretical maximum of dial-up is 56.6 Kbps - your local loop had to be blessed by the Gods "back in the day" to get exactly 56.6 Kbps. Getting 20 Mbps is pretty darn good.
Also, that first-hop of 10ms is likely your ISP's B-RAS - your Layer 3 gateway to the Internet, usually deep inside the CO and still a couple hops away from the Internet. Again, you have exceptional service if the B-RAS is 10ms away, and Internet only 17ms. The most common I see in Canada is 20ms/30ms, respectively.
There are lots of opportunities for network hardware to introduce needless latency:
Overview:
http://www.bufferbloat.net/projects/bloat/wiki/Bufferbloat
Demo:
http://www.youtube.com/watch?v=npiG7EBzHOU
Who cares about their test servers, what's the latency to your DG. If that's low then it's not a problem with your modem, it's hops within their network. Try finding an IP in your closest major city and run a trace to it. You'll be able to see where traffic is bottlenecked. Also, check the latency to your DNS. That can make a huge difference in surfing experience. I've seen some bag legs (like out in the country) that consistently test at or around 100ms. Anything more than that and I'd bitch. Voip starts getting choppy, games lag, etc.
See pic: http://www.ko4bb.com/Cox_Latency_Feb_2012.png
Embarq is on DSL and they should do much better than that. Particularly they should *consistently* do better than that.
It's time for another ISP. I had Embarq for phone service until I realized how much I was overpaying for substandard service. Even cancelling the service was a stressful experience. A month after cancelling and being told I would get a partial refund, I received another bill instead of a refund.
I have a 1.5 Mbs DSL line to Cavalier telephone. Pretty slow throughput-wise, but my pings to 8.8.8.8 are 10ms.
It's pretty good for gaming.
My phone does 100ms via UMTS or HSDPA. My cable connection at home is stable at 60-75ms typically, and I hate my modem for somany other reasons.
125ms is below what should be a standard. Rotsa ruck getting it fixed.
deleting the extra space after periods so i can stay relevant, yeah.
When I had Speakeasy (RIP) I had latencies of around 10ms. With Century link they have been anything form 40ms to 200ms at various times. And this was to the next hop after my modem. I consider latencies about 20ms to be annoying. 100ms is way too high for the hop from my modem to their router.
Hops after that are really hard to come up with values for since there is so much that can affect latency. But IMHO, if the broadband provider can't give you a link with less than 20ms latency from your modem to the next hop in line, something is horribly wrong with how they're managing their network.
I even question 10ms. In reality, it should take less than 5ms to get a packet anywhere within a metro area.
Need a Python, C++, Unix, Linux develop
http://www.linuxpromagazine.com/Issues/2011/127/Security-Lessons-Bufferbloat/%28kategorie%29/0
In this article, I’m not going to talk about an emerging technology (don’t get me wrong, I love new technology) but about something even more interesting: An emergent behavior that was never expected: bufferbloat.
Bufferbloat is not a recent phenomenon; however, it has only recently been uncovered and understood, and developers will likely be grappling with it for some time. Additionally, this problem, if left unchecked, will make the Internet painfully slow to use, greatly reducing the availability of services. Remember, availability is one of the three legs of the AIC triad (along with integrity and confidentiality).
So when people say "congestion causes slow networks" they are quite often right, but not for the reasons they think they are. Case in point: my Cablemodem ping times to www.seifried.org are nice and fast, until I saturate my uplink (with even just a single upload stream) at which point the latency increases to one second (in a semi-linear fashion over a few seconds, you can almost hear all the buffers getting filled up along the way).
He's spot on. The other question is how you're measuring the latency - lots of systems place a low priority on responding to pings, for instance.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
Latency that high is unacceptable - you can't use an IP phone with that kind of lag. You can't play games online, and I seriously doubt you could watch youtube with it. I'd call in again and escalate the issue. If they can deliver that with their contract, you're screwed.
Why are so many of you posting like you know what you are talking about when you clearly don't know the difference and relationship between latency and bandwidth.
The Kruger Dunning explains most post on
If you mean "ping response time", which is a good approximation of "round trip delay" for travelling back and forth to the servers AND including additional time for servers to respond, your number may be good.
May, if servers add delay. Answering pings is not a priority of a networked device. That's why "ping" is not that much useful except for verifying if a host/server is alive.
If this is only network part, one-way delay of 300ms would be excessive. Even 150ms (assuming original figure was two-way) seems a bit high. In USA you should get about 50-100ms from coast to coast inside telco network, adding under 50ms for your access link (higher for satellite or mobile, obviously).
For DSL your local link latency may be increased due to interleaving and other advanced encoding to cope with the length of copper wire you are on (if far from DSL concentrator). This is fairly low, on order of 30ms though.
(Note: this is of course a huge simplification, so please do not point out "mistakes", but add more information as you feel necessary)
Gaming (ETQW) from the Greater Washington, DC metro area to servers in Dallas or Chicago I see 50-80 ms times. To servers in Germany and France, 130-170 ms. Russia (Moscow), 150-190. Anything over that guarantees crappy game play and getting owned by the noobiest of noobs.
Learning HOW to think is more important than learning WHAT to think.
I can send a test ping from my home near the east coast all the way to a Sonus Soft switch in Los Angeles and back in 65ms. Back in the dial-up days it was acceptable to have 100ms ping times across a single link. A T1 should have ping times in 15ms across the single hop. DSL should be in the 50ms range on a bad day.
I had a client with a similar issue. Turned out there was a phone in the basement he had forgotten about and did not have a filter. Installed the filter and he got between 60 - 100ms to his default gateway where before it was 250+ ms
I lived in a rather not-so-large city in the south of Argentina in 2005, when ADSL was just starting to become available to middle class (at least there), and 150ms was the norm (I remember this because a lot of friends would play CS online with people from the rest of the country).
I've created and run a number of global networks and 300ms would be high if your ISP was half-way around the world from you. Unless they were based in a country with bad infrastructure like rural Vietnam (even then we averaged 200-280ms from the US). I find anything over 30-40ms to my ISP to be a bit lame since it is more than reasonable. I'd say depending on the type of connectivity that a threshold of 80ms is fair. 100ms is on the upper side of ridiculous if this is the average latency.
There are a lot of factors though, it could be some QOS or packet shaping/inspection that is the cause of high latency, it could be poorly configured gear at one or more hops, it could be a myriad of things. Personally, I would not accept anywhere near 300ms as my average latency.
http://teasphere.wordpress.com - A little spot of tea
I would be interested in a few pieces of information before attempting to judge the latency. First, what is your distance from your DSLAM/SLICK/remote? If you are on a 10 meg connection on DSL (are you actually on ADSL or a different service?) then 300MS latency to pretty much anything should be a red flag. Are you testing on a clean machine? Malware can consume massive amounts of bandwidth silently. Can you isolate the network and test directly through the ADSL router/modem? Again, if this is actually ADSL, do you use the circuit for voice communications as well or is it only in place for broadband? Is the voice quality acceptable? If it is only in place for broadband, what are the noise margin readings and what is your attenuation? If you are on a clean network (no malware, one PC directly connected to router), with no malfunctioning hardware on your side, and testing directly to one of your provider's servers/routers, then a 300ms ping time would definitely constitute a problem (at least by the standards of the company for which I provide support) and would warrant a technician further investigating the trouble.
So long as you are told what to expect when you sign up for it.
Even a latency of a second is not horrible for many purposes, while others need under a couple hundred ms to even be usable.
Troll is not a replacement for I disagree.
When your ISP has to keep changing names because they keep destroying their brand (USWest->Qwest->CenturyLink), that should tell you something right there.
My completely subjective experience over a decade of broadband use is that when things are working normally I expect less than 50ms for ADSL, less than 40ms for Cable and less than 20ms for Fiber, within national borders in a European country.
https://dalgamotor.wordpress.com/ - Elektronik beyinlere ozgurluk asisi (Turkish)
The first question I'd have about high latency numbers is how you're measuring them. Lots of devices are pretty slow about responding to pings and traceroutes. (Big routers, in particular, tend to make that a much lower priority than routing packets or doing other useful work, and the ping response comes from the CPU. while the actual packet routing happens in ASICs.) On the other hand, doing a traceroute to some distant site can let you see a bunch of dubious measurements, and the smallest numbers tell you a lot because they're a ceiling on the latency of everything up to that point. I've also seen throughput measurement tools that think sending 18000-byte pings is a good idea, and they're not only hopelessly broken for measuring throughput, they get really entertaining latency results as well. The quick and dirty test is "ping 8.8.8.8" followed by "traceroute 8.8.8.8", which points you to Google's anycasted DNS servers.
Traceroute also gives you some hints about routing - if you're in San Francisco, and your route to google.com is going by way of New York, something's weird with your ISP's peering. (I've seen that kind of thing happen - the user's ISP in Denver had recently moved, so their upstream link to the Tier 1 the user's headquarters used was down for a couple of months until they got a bigger access line built to the new site, and their ISP's other Tier 1 upstream didn't peer with the first Tier1 in Denver, and the San Francisco peering was overloaded back then so they were getting routed somewhere awkwardly far away.) But even so, it's really hard to burn more than an extra 120ms with bad routing unless you cross an ocean. (That's two extra round-trips across North America, or dancing around Europe; Asian users can occasionally get weird routes.)
The next thing to do is be sure you're really really not running anything else while running your latency tests. Jim Gettys's "Bufferbloat" paper is really insightful, and you need to read it (but don't measure your latency while you're downloading it :-) A typical latency problem is that you're trying to download more bandwidth than something on your access line can support (such as your wifi router), so the device buffers traffic, and what you're really seeing is that bittorrent or big http transfer is filling up your wifi to maximize throughput, which is trashing your latency. Or alternatively, you've got something hogging your upstream, making it difficult for ACKs on downstream traffic to get through.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
Of course, it depends on where you're pinging ... ... here's what I have on a residential line in San Diego ... to Los Angeles: 12ms ... to Yahoo: 57ms ... to London: 164ms ... to 8.8.8.8: 65ms
If I had 200ms, I would assume something was wrong ... bad wiring, interference, or a congested/broken switch enroute. Try tracert to find out where the delay is.
I'd be surprised if they sent a tech out to the site rather than testing remotely, but 200ms to anywhere in North America certainly indicates that either there's a problem or you're testing to something that doesn't respond very fast. Usually a high latency indicates that there's a buffer somewhere that's full, either because you're trying to push more traffic through it than it has room for right now (go read Gettys's Bufferbloat paper), or that there's a piece of equipment that's not working right (so "more traffic than it has room for right now" is much different than "more traffic than it's supposed to have room for), or something's misconfigured (like an Ethernet that ended up at 10mbps half-duplex instead of 100 or 1Gig full-duplex.) If there's wifi anywhere in the path, maybe it's getting a low speed connection or interference, so your 10 Mbps DSL or 60 Mbps cable modem is cramming bits toward a 2 Mbps wifi, or you're losing enough packets that you're getting TCP timeouts and retransmission, etc. Or maybe your copper access line is bad - I remember having similar discussions a few decades ago when I could only get 1200 baud instead of 2400 on my modem line, because the phone wire was rubbing against a tree branch.
But if the problem's anywhere upstream from your nearest network box, it usually means that either something's broken (and maybe you're the first person complaining), or there's a capacity planning problem (rare for the wired part of the network - they usually overprovision because it's simpler and cheaper than troubleshooting, but occasionally they'll do load balancing tricks instead), or something's gone weird with routing.
Bill Stewart
New Fast-Compression-only CPR http://preview.tinyurl.com/dy575ks
It depends on your ISP for their latency limits, I used to work for Comcast for 5 years and I host various dedicated gaming servers from home and I can tell you with certainty that if pings got over 150ms on the Comcast side to your modem (which is usually a ping from the local/regional office to your modem) you could request a tech (which they cant refuse) as they can adjust settings at the poll or node to drop those times (to a point) as its usually a filter gone/going bad. It also depends on your line quality (I've seen some bad ones that they couldn't do anything about even replacing the entire line up to the node), other times it's a modem problem (the scientific atlanta 2100/webstar and Terayon modems were notorious). It could be buffer bloat as others suggested above but I would give the ISP a call and request a tech out first and do the easy fix before messing with your own hardware, especially if that same hardware never had any issues with other ISPs.
The "test servers" you mention could be any number of hops away, so I wouldn't consider them a reliable way to test latency. I always test latency by pinging my ISP's closest router pointed to by my default route. For me, anything over 20ms would be unacceptable. I am usually able to find game servers at around 50ms.
If you have to ask this question, you probably aren't doing anything too serious, but personally, I need a fast connection to do my job. Right now I'm getting 9ms to about 30 miles away. That's acceptable to me. I don't know how much it would have to go up before it became unacceptable, but I know I would be complaining well before 300ms.
It's obviously bothering you, so I would say it is unacceptable. Start threatening to take your business elsewhere. If there is no alternative ISP in your area, get your neighbors to complain too. If they still won't take action, you and your neighbors call up an ISP available in a neighboring area and say that your ISP is not fulfilling your needs, so you hope that they become available in your area soon.
Really, though, if you were only told "300ms latency (to their test servers) is the acceptable range for Centurylink 10.0Mbps" by one person, take it with a grain of salt. ISP support people will say anything they think they can get away with to get a customer to stop complaining when they can't or don't know how to fix a problem.
It takes at least a third of a second to load a web page if all resources but the main HTML document are cached. If you have to load additional pictures and scripts, it will most likely take more than a second, just due to latency.
Typing in a SSH terminal will appear choppy and using the interactive functions of the sheell (arrow up, down, etc) will become uncomfortable
The delay in VoIP is clearly noticeable, but it's not impossible to have a conversation
Email is completely unaffected, or like browsing if using webmail
Large downloads and torrents are not directly affected by latency
Somewhat bad lag in most games
Of course, you probably know this, and was asking about what is normal. Here's another data point: 5Mbit DSL, 67 ms to my provider's website with a couple of torrents running
I'm participating in the European broadband performance monitoring trial going on right now across Europe. I have a little white box hooked up that runs ADSL line tests during off-peak times and when my line is not in use. I have a 24 Mbps down / 1 Mbps up connection to my ISP, and the latency average (taken from my monthly report, not made up) over the past month was 29 ms. Now, this is not the latency to my ISP, it's the latency to test servers somewhere in Europe, I really do not know where and haven't been bothered enough to dig. Considering I live in an historic neighborhood with old copper, I think that's pretty decent. I don't quite reach the throughput marketing numbers, but I do get decent speeds.
Whenever I visit family in the US, I notice the latency seems to be much higher there. I wonder if some or all ISP's intentionally insert delays to pump up latency and make the speed feel really low, because my parent's ISP calls them regularly to ask if they would like to upgrade their 20 Mbps service, which for reading email and light surfing should be WAY more than enough. Hell, I got by just fine ~12 years ago on a 128 kbps ISDN line for that stuff.
Your problem is CenturyLink. They bought-out verizon at my parents home and became the only available provider. Though my parents pay for 12Mbps down, they get 2Mbps. We've seen 4 different techs come out 6 times and try and fix the problem. 1/2 the time they've come out, said they were gonna do some tests, and then left saying it's fixed. The latest tech actually ran a line across the neighbors grass and into their outlet. Needless to say, the problem isn't fixed, the solution is incredibly stupid, and I can't wait for the neighbors to start mowing soon and rip the wire apart.
Century link says if they can get my parents to 3Mbps then they are meeting their minimum standard for that package. They aren't willing to try and get even close to the 12 advertised. The problem is squarely with their equipment as previously my parents got 12 down perfectly fine when they were paying verizon.
If you can, run from CenturyLink.
I would be rather worried if my ADSL connection got past 50 ms for the first hop after the router/modem, but that said, the real clincher (at least to me) is Voice over IP like SIP phones and, well, Skype -- a latency beyond about 150 ms end-to-end makes it difficult to "duplex" properly (i.e. interrupt the other without too much frustration) and 300 ms is near the edge of tolerable.
;-), the answer is: Less is more!
For gaming (and real-time financial transactions
At the end of the day, 300 MS latency can either be perfectly acceptable, or absolutely horrendous. If you're playing an online game, 300 MS latency means that there's more than a half second delay between when something happens server side, and when you react to it. That's enough to get you killed in an FPS or MMO, even though half a second, in theory, doesn't sound like all that long. That's absolutely unacceptable. If you have to wait half a second before a web page starts loading or a download begins, that's a different story altogether. It also depends on what's causing the latency, and whether or not that level of latency is a regular occurrence or an unusual circumstance. From the sound of it, this has been a month that you've been seeking resolution, so it sounds like this is a regular occurrence. How are you measuring the latency? Is there a particular site / game / etc that it's this high for, and not necessarily all the time? Are you using a wireless network, or wired? What is your latency to the router? If it is all the time, and if you're using it for gaming, or other situations in which higher levels of latency are severely impacting you, it's a good time to shop for a different ISP. FWIW, I usually sit, give or take, around 20 ms latency with FIOS. There have been spikes of much, much worse, sometimes for as long as a day or two, but that hasn't been for quite a while, and has never been sustained long-term.
At one point, I was in the top 50 or so Tac Ops players in the world. Anything over 50ms and the aim bot users always win. Between 30 and 50 I could compete effectively against the aim bot users. I don't know if I would consistently surpass the aim bot users at under 30ms as I never had the opportunity to test that theory.
I'm centrally located in the continental U.S. I would consider anything over 100ms to a destination in the continental U.S. to be unsatisfactory. More than average latency, though, I'm concerned with the variability of latency. Spikes = bad, bad, bad.
I get 4 ms to the closest SpeedTest.net server, which is not on my ISP network, but I'm on a 100mbps FTTH connection. I'd consider up to 20 ms to a local server acceptable, but that's the limit. When I still had cable, I remember gaming on UK servers with 30ms latencies, and that's 1500 Km away.
If ping to my ISP is more than 20 ms, I am not happy. I play online fighting games and they require good latency. In fact, even the speed of light is too low to play against people on the other side of the planet.
If my ping to most places within a reasonable range (2000 miles) is under 100 ms, it is acceptable.
50 ms is good. 30 ms is perfection.
For me, the 2 reasons I need good latency are online gaming and, to a lesser extent, Skype.
I have had Cox Cable for my internet for about 5 years. Its the only service available in my area (Phoenix, AZ). I've noticed over that time the average latency for gaming servers has gone from about 35ms to 150ms. The gaming experience has noticeably suffered correspondingly.
Over the years, I tried complaining to Cox many times but their monkeys only know to run some remote test to my cable modem then tell me my connection is fine. They often suggest I upgrade my package to a higher bandwidth one. I did actually try that for a while. It certainly improved the bandwidth but the latency was the same. They also suggested I upgraded my cable modem to a Docsys3-capable one. I did. It made no difference whatsoever in latency or bandwidth that I could see.
Anything over 50ms round trip to a device on your ISP's network is a problem. I get 34ms consistently to the Verizon FIOS test server (speedtest.verizon.net) for example. Leaving the ISP's network, though.. all bets are off.
I've got four isp's to choose from for broadband where I live fortunately (not counting pseudo broadband like satellite and dsl), so if they had a problem and wouldn't resolve it, I'd just switch. You might not be so fortunate, I guess.
is it just me, or is anything above 25ms ping time suspect? i'm on a 30mbps Comcast connection, and i've always got sub-20ms ping time (according to speedtest.net). 300ms sounds unacceptably crazy high to me.
First, just to be clear, the only thing you can hold your ISP truly accountable for is what you are actually paying for, which is your first HP latency, that is from the point the packet leaves your local network and enters their network.
Typical latency by connection type:
Ethernet:.3ms
Analog Modem:100-200ms
ISDN:15-30ms
DSL/Cable:10-30ms
Stationary Satellite:>500ms, mostly due to high orbital elevation
DS1/T1:2-5ms
I have had DSL for about 9 years. I have had MRTG monitoring my link to the local DSL router, the remote DSL end point, ATT.com and dslreports.com for pretty much all of that 9 years.
For the last year these are my stats to the far side of my DSL link:
(Max/Average/Current )
High: 957.0 ms 28.0 ms 27.0 ms
Low: 907.0 ms 27.0 ms 25.0 ms
I am 1127feet to my C/O which is about as good as it gets. Anything over 30-35ms and I would be raising holy hell.
In the case of a two-way satellite system, when you request something by clicking on a link, or any other way, that message travels 44,600 miles just to get to the NOC. The stuff coming back to you must travel the reverse route, so the round trip is 89,200 miles.
The speed of light is 186,000 miles per second in a VACUUM,slower through the atmosphere. But even if you assumed 186,000 mps then the total time taken in space travel is about 480ms. Given the atmosphere problem, it is actually more like 500ms. Add to that the terrestrial internet latency, which should be about 100ms. Also you can add delays through transponders, gateways, proxies, etc.
At 300ms your packets are going to a NOC to a Satellite, back and to another NOC for your first hop.
You can see ATT to CenturyLink Latency here and their current average is ~35ms
here.
Armaments, 2-9-21 And Saint Attila raised the hand grenade up on high, saying, 'O Lord, bless this Thy hand grenade' N
I would accept 150+ ms if the ISP was on the other side of the Atlantic. Anything above 30ms to nearest Internet exchange center I'd report as an error to the ISP.
A normal coast to coast connection in the USA(East to West or West to East) will be upwards of 85ms, and 150ms or so is fairly normal connecting to a server on another ISP(multiple ISPs in the middle). Anything above 300ms or so to just about anywhere in the USA indicates some sort of problem between you and that other system.
Anything above 100ms and I begin troubleshooting. I normally get about 70 ms from my CenturyLink connection on a 5-7Mbps line (supposed to be 7 but I normally bounce between 5-7).
I usually deal with this kind of problem by switching to IP over Avian Carriers and pinging their headquarters to death!
As a frame of reference, I'm also with Century "NO" Link with 10mbps DSL, and pretty rural. Indeed, if I had ANY other viable choice, I would leave CL in a heartbeat, as their service is terrible. However, my latency is typically around the 30-40ms range when I am connected, so if you're getting ~300ms, then there is an issue somewhere. Given my history with them, most likely the problem is on their end. My neighbor is with them also, and also has the 10mbps package, and we will both simply lose connection at random times throughout the day, almost every day. I've had their techs out many times over the years, and the problem still exists. In fact, just this last Friday it was down for more than 4 hours. Sometimes it's 4 seconds, sometimes 4 minutes... unfortunately, sometimes it's 4 hours too. I call every time there is a disconnect, yet they can't seem to find an answer. They're just awesome... or not. Switch providers if you can!!!
But then QWorst/Centurylink has dropped my connection four times in the last nine hours. Not that unusual days after a storm. If I didn't have _so_ many reasons I don't want Comcast.....
Getting 20Mbps from ADSL2+ pretty much means you're living across the street from your CO
I've got better things to do tonight than die.
For each type of network, there's a latency per hop and a (longer) latency at the endpoints for packet processing. (Pingpong tests generally use zero payload, so will just give you latency per hop.)
At the hardware level, cheap telecos might connect up using T2 or T4 lines (these are teleco-to-teleco lines, you can't buy them). These will have very different latencies to inter-teleco SDSL or inter-teleco optic fibre.
The on-the-wire protocol also matters. There's going to be a big difference between using ATM, Frame Relay or SONET, for example. On top of the on-the-wire protocol, there are next-level protocols that also make a big difference - MPLS, for example.
Finally, there is the topology. Many cheap telecos use tree topologies (which minimize their costs but maximize your hop count), some use simple mesh topologies (typically shortens distances to about half those in a tree), a few might use more complex meshes (reduces hop counts still further) or other topologies.
The difference in performance overall between the best-case and worst-case can be two or three orders of magnitude. For moderate improvements in performance, the difference in cost will be a fraction of the difference in improvement in service. For major improvements in improvements in performance, you will pay through the nose.
There is no -theoretical- reason why you could not achieve 8ms per hop latencies. The WAN hardware for this exists. There is no -theoretical- reason why backbone providers couldn't handle 10 terabit/second rates per fibre - again the hardware exists. There is no -theoretical- reason why you could not get 1 gigabit to the home - it's already done in some cities in some countries. In practice, private ISPs are run for the benefit of their shareholders, NOT for the benefit of the consumer. They will deliver only what they absolutely have to in order to keep the profits up.
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
As that's what I've had since I went ADSL2+. Any significant increase in this will mean my ISP will hound the telco until they fix it, because they actually give a shit.
I'm also a Century Link customer, in a small town in a rural area. My speed is 4 Mbps, although I'm paying for more, they have to cap me at 4 because my distance from their station won't work reliably at faster speeds. My ping times to 8.8.8.8 are running right at 50 ms:
Pinging 8.8.8.8 with 32 bytes of data:
Reply from 8.8.8.8: bytes=32 time=50ms TTL=54
Reply from 8.8.8.8: bytes=32 time=50ms TTL=54
Reply from 8.8.8.8: bytes=32 time=48ms TTL=54
Reply from 8.8.8.8: bytes=32 time=49ms TTL=54
Tracert:
1 4 ms 1 ms 1 ms 192.168.6.1
2 2 ms 2 ms 3 ms 192.168.2.1
3 32 ms 31 ms 31 ms va-71-51-24-1.dhcp.embarqhsd.net [71.51.24.1]
4 31 ms 32 ms 31 ms va-71-0-83-121.dyn.embarqhsd.net [71.0.83.121]
5 42 ms 44 ms 46 ms 208-110-248-237.centurylink.net [208.110.248.237]
6 58 ms 57 ms 55 ms 72.14.219.254
7 50 ms 50 ms 50 ms 209.85.252.46
8 42 ms 43 ms 44 ms 64.233.175.109
9 52 ms 56 ms 50 ms 72.14.232.25
10 50 ms 50 ms 50 ms google-public-dns-a.google.com [8.8.8.8]
Maybe that can be of some use to you in your complaining to Century Link and what is "possible" or "normal" for other customers.
Better known as 318230.
I get about 20 ms on a 6 Mbps ADSL link.
Votator.com implements a fair voting scheme (free
Actually it was the max throughput of a 24 channel channelized T1 at 64Kbit/channel (64Kbps * 24 = 1.536Mbps T1/ 24 channels)=64Kbps-8Kbps per channel. The overhead necessary to frame a T1 is 8Kbps. If you channelize all 24 channels, you lose 8Kbps per channel. Geeks were so anal about the reality of performance they wouldn't let you get away with the current bandwidth blasphemy we have today with DSL so they called a 64kbps modem a 56.6kbps modem because that was the best you would ever really get from it.
Armaments, 2-9-21 And Saint Attila raised the hand grenade up on high, saying, 'O Lord, bless this Thy hand grenade' N