is very dependent on the application. In particular it depends on the mix of linear vs random operations and the mix of reads and writes.
For 100% read applications SSDs tend to be similar in performance to hard disks when reading linearly, and a lot faster than hard disks when reading randomly. This shows up in linear read speeds of 100 MB/sec for a typical Flash SSD which is "close" to a hard disk. For random 4K reads, Flash SSDs can stomp any hard disk. Most disks are in the 10,000 4K read IOPS range where 15K SAS drives are in the 250 range or 40x slower. So for applications that are 100% read SSDs can be as much as 40x faster, although the average is usually in the range of 15x to 20x.
When you start writing to Flash things get interesting. Flash is really designed for large, linear, aligned, writes. With most drives, you can get maximum write throughput only if you write exactly aligned with the drives internal erase blocks. Thus you can write exactly 2 megabytes on exact 2 megabyte drive boundaries and get 100% of the theoretical write throughput of the drive. Unfortunately, no application acts like this, so you are at the mercy of the file system and Flash controller to turn your smaller, probably random, and probably mis-aligned writes into what the drive can handle. The net impact of this is that good Flash SSDs have 4K random write IOPS in the 120s which is 1/2 the speed of a 15K SAS drive. I have measured Flash SSD with 4K write IOPS with values like 135, 120, 64, 43, 24, 13, 4.0, and 3.3.
This is why Flash SSD performance is so hard to judge. The random write performance can suck up the available "drive time" and dig a system deep into dirty buffer flushing. We talked with one Dell laptop user that described their system becoming "unusable" while an Outlook indexing operations was randomly updating a big file. Unusable in this case was 2+ minutes for to bring up task manager.
These random writes also have a real impact on the wear of the drive. Every time you seek a write, you basically chew up a write/erase cycle, even if the write is only 4K long. If you look at a drive that claims 50 GB/day for 10 years, this is 50 GB of linear writes on exact erase block boundaries. If you write 4K randomly, the 50 GB really means 25,000 4K writes or 100 Megabytes of random writes.
The solution to this is to not write randomly to the drive. There are file systems designed for Flash that address these issues. These are typically called "Log File Systems". Unfortunately, there is no generally available file system really designed for performance. In Linux the LogFS options are really tuned for small memory small storage systems and for hardware where the flash chips are directly accessible. They do help drive wear a lot, but they are just not tuned for Gigabytes of space or database crunching performance.
Another solution is my companies product called MFT (Managed Flash Technology) which is a software block mapping layer that runs on the host. It gives you the random write performance benefits and wear benefits of a LogFS while allowing you to use whatever file system you wish. MFT was developed on 2.6 Linux and has been ported to Windows. With MFT, the same drives that do 25 4K random write IOPS usually measure over 10,000. The linear speed of the drive is still equal to a hard disk, but the random speed is now closer to symmetric with reads and writes. Thus jobs like updating databases can literally run 20x faster than the fastest hard disks.
In the end, Flash SSDs will find specific markets initially. I can say with certainty that they won't get used for off-line backups or storing/edit large quantities of HD video. But give them databases or file systems with lots of small files, and they can really smoke a hard drive.
Most current Flash SSDs have a 2 Megabyte erase block. Take a 32 GB MLC drive with 10,000 cycles and you wear out in 160 million random write operations. At 20/sec this is 92 days. At 200/sec it is 9 days.
Good enough for laptops. Not good enough for servers.
On June 15th, Mtron will start shipping the 1000 series MLC drives. Put these in an array with the right software and you end up with price/GB parity with 36GB 15K 2.5" SAS drives and about 12x the random IO performance.
HDD Array:
8 Seagate Savvio 2.5" HDDs: $350ea $2,800
configured raid-10 1 SAS raid controller $600 Total cost for 144 GB $3,400 or $23.61/GB
SSD Array:
6 Mtron 1025-32 2.5" SSDs: $290ea $1,740
configured raid-5 1 SATA raid controller $250 MFT Software License $1,250 Total Cost for 144 GB $3,240 or $22.50/GB
HDD Performance:
4K and 8K read IOPS: 250/2000 (single-threaded/multi-threaded)
4K and 8K write IOPS: 1200
These performance numbers are with the MFT driver in place. Without MFT, the 4K random write performance is about 140 IOPS (>250x slower).
Endurance for these SSDs in this configuration is good enough to overwrite the entire array with random data three times a day (500GB of random updates/day) for about five years.
These drives make a wicked mail server (EasyCo just moved one of it's mail servers mirrored to MLC flash and the difference is amazing).
Sorry for the blatant advert, but SSDs are here now.
I have build a couple of these for colos, but they are low cost, quiet, and low power, so you might be interested.
Start with a Supermicro 1U 510. This is 11" deep with an 80+ efficiency 200W power supply. It will house most uATX motherboards, although you will have to nibble out the back panel to get non supermicro boards to fit.
Then add a low cost AMD socket-AM2 motherboard.
Put a low power 45W dual core like a BE-2400 or one of the newer series. You will need a 1U blower for this.
Put one 3.5" drive (or up to 4 2.5" drives).
With a Gigabyte board and 1 2.5" HDD + 1 2.5" SSD, my system draws 37 watts in use as measured from the power line. It is not 100% quiet, but is the quietest of any 1U I have worked with.
My apologies for a long post. There will be some adverts embedded, but I will try to keep things informative.
The reason that Flash SSDs act "wierd" in benchmarks is that they have asymmetric performance patterns when reading and writing. Particularly with random operations, this asymmetry is huge. Here are a couple of example "drives":
* Mtron 7000 series: >14,000 4K random reads. ~130 4K random writes. * SanDisk 5000 series: ~7,000 4K random reads. 13 4K random writes. * Cheap CF card or USB stick: ~2,500 4K random reads. 3.3 4K random writes.
This is a 100:1 performance deficit when doing random writes versus the random reads. This has some really weird impacts on system performance. For example, if you run Outlook and tell it to "index" your system, it will build a 1-4 GB index file in-place with 100% random writes. If you do this on a hard disk, the job takes a long time and drags down your laptop, but the operation is still pretty smooth. Do the same think on an SSD and the system slugs to molasses. One of our customers described it as "totally unusable" with 2+ minutes to bring up task manager. What happens is that the fast reads allow the application to dirty write buffer faster and this then swamps system RAM, you get a 100+ deep write queue (at 13/sec), and you want to throw the machine off of a bridge.
This fix as some have described it is not some magic new controller glue or putting the flash closer to the CPU. It is organizing the write patterns to more closely match what the Flash chips are good at. Numerous embedded file systems like JFFS do this, but they are really designed for very small devices and are more concerned with wear and lifespan issue than performance.
Now here comes the advert (flames welcome). A little over 2 years ago, I wrote a "block translation" layer for use with Flash storage devices. It is somewhat similar to a LogFS, but it is not really a file system and it does not play be all of the rules of a LogFS. It does however remap blocks and linearize writes. Thus it plays well with Flash. It also appears to be an "invention", and thus my patent lawyer is well paid.
The working name of the driver layer itself is "Fast Block Device" (fbd) and the marketing name is "Manged Flash Technology". And what this does is to transparently map one block device into another view. You can then put whatever file system you want into the mix.
In terms of performance, it is all about bandwidth. Build a little raid-5 array with 4 Mtron drives and you will get over 200 MB/sec of sustained write throughput. With MFT in place, this directly translates into 50,000 4K random writes/sec. Even better, you tend to end up with something that is much closer to symmetric in terms of random read/write performance.
MFT is production on Linux (it has actually been shipping since last summer) and is in Beta test on Windows. It works with single drives as well as small to medium sized arrays. It does work with large arrays, but the controllers don't tend to keep up with the drives, so large arrays are useful for capacity but don't really help performance a lot. Once you get to 50,000 IOPS it is hard for the controllers to go much faster.
Consumer testing with MFT tends to produce some laughable results. We ran PCMark05's disk test on it and produced numbers in the 250K range. This was with a single Mtron 3025. Our code is fast, but we fooled the benchmark in this case.
There are several white papers on MFT posted in the news link of our website:
Swap to Flash is "stupid", but the stupidest in 55 years. Surely you can come up with something worse.
Anyways, swap of any kind only make sense if adding memory is inconvenient. It all depends on whether system memory is even available at the size you need.
Right now, RAM prices are so low that flash as swap does not make much sense.
Flash devices are fast at small reads but very slow at small random writes. Swap is usually a 50-50 mix of reads and writes with no pattern to be seen. Thus if you swap to Flash, at least traditional Flash, then you will be very unhappy.
Your comment about bandwidth needing to go to a spindle seems strange as well. I have a 4 drive raid-5 flash array here that just tested at >400 MB/sec on reads and >150 MB/sec on random writes...
I don't know of many "spindles" that can keep up with that. Plus total power draw is 12 watts operating and 2 watts idle (actually the raid controller draws another 15, sorry).
If you want to see how Flash "can" be used for swap, see:
Random write performance to bare drives is usually quite bad. Most "reputable" vendors do publish random write figures. SanDisk quotes 13 IOPS in the spec sheets. Mtron quotes 120 IOPS. I have not seen quotes from Samsung, but have tested their old drives at 27 IOPS. I even tested one drive at 3.3 write IOPS.
On the other hand, random writes issues are "fixable". My company just published tests for various Raid-5 Flash SSDs setups. For 4 drives testing with 10 threads on Linux 2.6.22 using our MFT "driver", we get:
4K random reads 39,689 IOPS 155 MB/sec 4K random writes 29,618 IOPS 115 MB/sec
These are real numbers and the application does see the performance improvement.
1 megabit of 95% percentile bandwidth is about 100G/month. This assumes you are "using about 30%" of the peaks on average. You might be doing a little better. At 2T, this implies that you are 95% peaking at 20megabits. The dirt-cheapest you can buy a megabit for is the low teens, so this would be $250/mo in raw bandwidth costs assuming you are buying a lot more than 20. If you bought this directly from a tier-1, 20 megabits would be about $70-$90/mbit depending on where you bought it.
The bottom line is that I do believe that you are running 2T for $120. You are probably renting a basic dedicated server from ServerBeach or similar and they bundle in a lot of BW with the deal. They are playing the averages. If you are actually using 2T, then they are losing money on you. This is why these companies won't sell you 4T for $240. They also won't sell you 2T with your server and often restrict porn. They really want the averages to run 100G/system.
DailyKos has >1M visitors/day. This is not page hits, but unique people. Do the math. 1M * 5 pages * 100K/page (they are big pages) = 500G/day or 15T/mo. This is ~150mbits bursing to ~400mbits, so we are talking a dedicated gig-e port. 200mbits of level-3 from colocation.com costs $50/mbit, so that is $10K for BW. It appears that Kos is not getting that bad of a deal after all (plus I suspect the $7K including things like racks and power).
Your getting 2T for $120 is akin to the Lieberman campaign. You get what you pay for.
I am happy to agree with you, but I suspect that I will draw the line at a quite different point than you might.
If you want to argue that a candidate cannot send you an email once a day (or once an hour) then I think you are wrong. If you want to argue that a candidate cannot send you 1000 emails per hour shutting down your inbox, then I think you are right. You really do have to go to extremes to find cases where restricting political speech is justified and constitutional.
If the speech itself makes your inbox useless because it crashed your mail server it is probably over the line. I would argue that it is mostly over the line because it would prevent other political speech from reaching you.
Regardless, a candidate making phone calls, to your cell phone, that is unlisted and on the do-not-call list, a couple of times before an election, is protected, even though it will bug the hell out of you. The alternative of restricting this is a cure that is worse than the disease. As email goes, this means that you have to put up with a dozen unsolicited emails from a candidate or advocacy group and it will bug you, but it is not reasonable (or smart) to ask for a law to prevent it. If you don't want to read the email, don't. Engage the sender. Campaign against them. But don't expect a law to protect you from being annoyed.
Remember, there is a constitutional right to free speech. There is no constitutional right to not be offended (or annoyed).
There is the obvious issue of "cost to deliver", but that does not change the speech. If you rent a mailbox at the local post office, junk mail and can does cost you real money. This is probably a lot more money than your email box.
If someone sends you an email, they can only send it to you because you have setup your mail server to receive it. They did not break into your computer system to deliver the mail to you. You setup a server that is designed to accept exactly what the spammer is sending to you. You might "prefer" that he not send this stuff to you, but you have not communicated this preference to the sender in any way, let along entered into a contract with him so that he agrees to pay you.
If your inbox is your property, then set it up so that it no longer accepts email from anyone. It is your property and you are prefectly within your rights to refuse to receive mail from anyone not on your white-list. If you don't setup mail this way, then this is your choice. If you want people to pay you for the priviledge of getting access to your email inbox, then setup your email inbox to send back a contract that the sender can sign. Just because you "want" something to work in a particular way does not mean that the other party agrees with you.
So you can use your inbox how you want to. You could have restricted access to it, but have chosen not to, or are just too lazy to want to go to the trouble. Your solution is to have the government jail some politician because he is somehow violating your property rights, even though you have done nothing to protect them. Your inbox is an open, green, field with no fence, no "do not trespass" signs. Nothing to indicate that everyone is not welcome. You have elevated your right to not be annoyed and minor incremental costs of running an inbox (of which 99.9% are the cost of your time, ie. your annoyance, and not the actual cost of the inbox) over the constitutionally protected right of free political speech.
If this were an UCE (Unsolicited Commercial Email) and not your expanded definiation that include political, religious, and other non-commercial email in the broad term of SPAM, then you might have a case. The constitutional protections on commercial speech are much lower. But this is political speech, so you are just plain wrong.
I get really amazed at the people that think that SPAM is equal to mass murder.
SPAM is many things to many people. To some, it is an email message, commercial in nature, and unsolicited. To some it is any unsolicited email. To others, it has to be sent in bulk (whatever that means).
Regardless, if the email itself is political in nature, it is covered by free speech. If you don't want to read it don't. If you want to filter it out fine. If you want to only get email from your frields, blacklist everyone else. Don't somehow claim that political emails are not free speech. They ABSOLUTELY are free speech. For you to argue otherwise is simply amazing.
And yes, a politician can come to your front door (during reasonable hours), ring the doorbell, and ask to talk to you even if you have a "post no bills" and "no solicitiing" sign prohibiting this. This is free speech. If you are registered to vote, the political parties and candidates can get your name and address and can contact you in whatever way they can figure out that does not cross the line into harassment and trespassing. You cannot opt out of this. If you want to opt out, then you cannot register to vote. Sorry. This is how the system works (at least in California). And this is a good thing because voting and political speech is far far more important than spam.
When I lived in Georgia a couple of decades ago, we would occasionally see the Klan going door to door and standing on the street corner. I did not agree with it, but it was free speech. It was disruptive. You could even argue that it was dangerous and evil. It was still free speech and as such protected.
Remember guys that getting an unsolicited email is not mass murder. In political terms, this is advocacy. You can complain that using email is "doing it on the cheap", but then again, giving lesser financed voices a stage is probably a good thing. Your power against this lies with shaping public opinion, voting against a bad policy, or advocating against the positions that you disagree with. To say that political speech does not belong in email is just unamerican (and I use that phrase intentionally).
Your logic is all wrong. You have setup a mailbox that accepts, from the public, inbound communications. This is no different than having a mailbox with a street address. By setting this up, you are accepting speech from the general public, including from those that are anonymous. You did this by your choice. Now you are saying that the government, using its ability to commit violence on its citizenship, should pass and enforce a law that restricts what someone can say to you, and moreso, what someone can say as political speach.
Receiving an email is not "taking over your living room". That is the stupidest analogy I have ever seen. Getting junk mail in your post office mailbox does not require you to listen. If you are renting a mailbox at the local post office, the anoyance factor of junk mail is about the same as filling your email inbox (and the actual hard cost is probably more). Free speech means that people are free to talk. It means that they are free to talk to you. It does not mean you have to listen, but it does mean that you have to put up with the annoyance of them "trying" to talk to you. Sorry, but you do not have a constitutional right to not be annoyed.
And I am very happy that you are not an elected official or judge with the responsibility of making laws or interpreting laws or the constitution. Free speech is much more important than your inbox. Sorry if you disagree with that.
But you are wrong. Email is absolutely speech. Your argument is that receiving an email is somehow trespassing. Unfortunately, you invited this trespass (or at least you did not take any action to prevent it). It was you and your ISP that made the decision to receive un-authenticated email from the public internet. If you don't like the email, filter it, don't read it, or just delete it. You are not required to read it anymore than you are required to read a billboard on a public sidewalk.
The real issue here is what is more important. Free, political, speech, or the convenience of your inbox being less cluttered. If you think your inbox is more important, then you just don't get it. Some things are truely more important than convenience and making political speech, even if it is annoying, illegal is a very bad idea and the definition of the "solution is worse than the problem".
The CanSpam law was already on somewhat shakey legal ground. In the US we have the 1st ammendment, you know, the one about free speech. Regardless of how much you might dislike it, spam is a form of speech. Usually, it is a form of commercial speech.
The interpretation of the constitution deals with speech issues differently depending on the type of speech involved. There is very little protection for fraudulent speech, so this can be made a crime and you can be put into jail. The same goes for yelling fire in a theater.
Commercial speech is somewhere in the middle and the government can regulate it, but only at the edges. In general, current constitutional interpretations allow for almost all commercial speech.
Political speech is at the other extreme. Whereas fraudulent speech can be outlawed and commercial speech can be limited, political speech is "the most protected" type of speech. Political speech might be annoying, might make you angry, and might even be racist or offend you. Regardless, if it is political speech, then it is allowed, pretty much without restriction. The constitution guarantees you the right to free speech. It does not guarantee you any right to not be offended or annoyed.
Political speech is also not limited to a candidate running for office. Groups often make unsolicited calls, ignoring the do-not-call lists, on behalf of an issue or cause.
Before you complain about this and suggest that "there needs to be a law", think very carefully. First, I doubt that a law would do it. The free speech protection is in the constitution so a law might not be valid anyway. Regardless, is your comfort at the dinner table or at your PC more important than the constitution. You always have the right to not answer the phone, email back to jerks that pollute your inbox, call them to complain, waste their time (while being careful to only express your opinion and not harrass). It would be a very bad idea to want the government to "pass a law" allowing the government to "thru the use of force" put someone in jail because of what they say, especially when the topic of the speech is political.
The bottom line here is that there are other ways to achieve balance. If a campaign uses email and does not manage their lists and remove requests well, don't vote for them. This is not a no vote because of disagreement with the candidates positions, it is a no vote because of a candidates lack of organization and poor adminstration of their data. If a candidate cannot properly handle an email list, do you realy expect them to be able to run a government office.
The other point here is that not voting for a candidate just because they sent you an email or called you with an automated message might be really stupid on your part. If you are too lazy (as many US voters are) to vote anyway (let along actually check out issues and learn about your local candidates), then perhaps you deserve to be interrupted at dinner. I find that the best way to turn a candidates machine off is to say "I have already voted by mail". Gets them every time.
Level-3 and Cogent are not all that differnet in size. They both cover the US from coast to coast. They both go to Europe. Level-3 goes to Hawaii, but that is the only real footprint difference.
Cogent has fewer US POPs, but still hits all four corners. Cogent claims 80G backbones. Level-3 claims 110G backbones. Both networks have real 24x7 operations centers and skilled operators.
You cannot deny that there is a differnce in size between the two companies, but it is pretty easy to describe both as "Tier-1" networks.
(which is where the Level-3 bw numbers come from, so they might be a little old), it would look on the surface that peering with Cogent is perfectly reasonable.
The complaint of Level-3 is twofold. The first one is, "we are bigger, so you should pay us". It is true that L3 is bigger, but whether they can throw that around and make Cogent pay is not 100% clear. When you buy from Level-3, you "expect" to be able to talk to Cogent. I am a Level-3 customer and L3 did not imform me that they were cutting these lines off. I was pissed. I knew that L3 could legally do what they wanted, but I thought it inappropriate for them to do it without giving their customers notice. Level-3 chopped off 45 million IP addresses when they de-peered Cogent. Even if L3 thought it was Cogents "fault", it was still L3 that threw the switch.
One other issue of this is that some major suppliers buy from L3 and don't multi-home. Thus, RoadRunner cable modems lost contact with all Cogent sites. This left RR answering the phones from customers. I am sure it cost RR a lot of hassle, and probably some real money. I cannot imaging that RR was not threatening L3 with "we are going to multi-home and you will lose 60%+ of our traffic if you don't get this working now". L3 had a lot to lose.
This is the real problem with Level-3 taking action. Whether Level-3 is right or wrong, it's customers will still blame L3. L3 did enter into peering with Cogent. L3 did initiate the termination of the peering arrangement. In doing so, L3 did cutoff a large chunk of the internet to it's own customers. Image if Level-3, on it's home page, prominently stated that you were not buying connection to the "whole" internet. No one would buy service from them. Level-3 gets customers because, on the whole, we trust Level-3 to work hard to keep us connected. This trust is what L3 is selling. The de-peering has damaged this trust and will, in the end, cost L3 a lot more money than they were trying to extract from Cogent.
The only real, technical issue between L3 and Cogent is traffic symmetry. Cogent sells to a lot of hosters. L3 sells to a lot of end-users (cable modems, dial-up, etc.). Thus more bytes move from Cogent to L3 than from L3 to Cogent. If the two networks only touched at one point, this would still be fair. After all, it takes a L3 customers to request a Cogent porn page. But the peering between L3 and cogent happens at multiple points. Thus, you get the "hot potato" case where traffic from Cogent dumps to L3 early (and vice versa, traffic from L3 dumps to Cogent early). This means more Cogent bytes travel on L3's network than on Cogents network. This is presumably what L3 wants Cogent to pay for. While this may be a "real" traffic issue, I suspect L3 wants more from Cogent than is actually reasonable. What would be interesting would be for L3 to advertise its routes to Cogent differently at each peering point. Thus Cogent would carry it's own outbound traffic over more of it's own network and the "hot potato" effect would not happen. Perhaps BGP does not work that way, but it should be able to. I suspect L3 was more interested in "monetizing" it's relationship than in making it "fair".
I also just noticed that L3 is starting to compete with Cogent on price. It used to be that the $3000/mo for a 100mbit
For every republican you can find that's corrupt I can find a democrat...
but then again, it took the dems 40 years to screw things up, and the repubs have done it in 10.
The problem is not who is dishonest, but who is dishonest and efficient. Repubs and dems are both dishonest (or have dishonest members among them), but the repubs are head and sholders above dems at being organized, zelous, dogmatic, controlling, self-rightous, etc.
And in talking about Clinton, I will take the midget of the cabinet (Robert Reich) over each and every official on the W team.
I don't know of an existing tool for this, but you want to measure "known traffic" across the route and report when it degrades.
Setup an "application ping server" on the far end. This will be a C program that posts a UDP listen on a port in the RTP range. When it gets an inbound packet, it returns exactly that same packet to whom sent it to them. This needs to be in C (or similar) because the latency needs to be very low. It should also run on a very low utilization server.
On the near end, write a similar program that sends a UDP packet to the far end and measures round-trip delay. If you want to really act like a VOIP call, send 200 byte packets at 50/sec (or whatever the SIP/RTP payload is for your particular codec).
The question is then how to analyze the data. I would probably have the program spit out peak/avg "numbers" for latency and packet loss into a text file, perhaps at 1/minute. You could then have MRTG (or your favorite grapher) read those and draw you pretty stuff. Once you see the types of degredation that actually matters, you can program in your thresholds to page/email/whatever you.
Things to remember when setting this up. Make sure the traffic is handled by the routers exactly like your VOIP packets. Run on the same UDP ports with the same QOS tags. In other terms, make your test traffic indistinguishable from the actual voice traffic (at least from a router's/switch's voicepoint).
Of course, someone probably has a package that does this. If not, build it and open a sourceforge project.
If I read this story correctly, the chain of events appears to "maybe" be a bit more interesting (but I am speculating on some of this).
* Borders was apparantly not paying sales tax (speculation) * They decided to pay the sales tax
* This might have been encourages by the California Tax
Amnesty program (again, speculation)
* ie, no penalties, but still interest
* Ahhhnolds plan to help balance the budget * Borders then sues the state to get the money back
* They lost * Borders then appealed
* They lost again (this is the current case)
In thinking of how borders runs their business, the line...
If it looks like a duck...
seems to apply.
If you don't like taxes, remember that for every tax that is avoided, there has to be another tax to make up for it.
I have a Verizon account. I have actually had it for a couple of years and it has improved markedly over time.
Verizon has two classes of available service. You can use a single PC-CARD adapter that will auto-switch depending on which service is available to you. There is even another/. article on the "Stompbox" Verizon to WiFi mobile hot-stop.
If you are lucky enough to be in a Verizon EVDO "Broadband" access area, you can expect 2 Megabits "peak" downstream speed. Not sure what the peak upstream speed is. Regardless, if I have reasonable single strength, I get about 500kbit down and 80kbit up. I think it is fair to call this "DSL speed". Even though the speed is good, the latency is still a bit high at about 200-250ms.
At this point, the "Broadband" access areas currently cover about 30 cities. If you are really "out in the woods", I would be very surprised if you are in one of these areas, but you can check. This is a list of the broadband access cities:
If you are not in one of the "Broadband" areas with EVDO coverage, you can still use the 1xRTT service throughout most of the rest of the country. 1xRTT gives you theoretical 144kbit down. I tend to get about 60-80 kbit down and 20kbit up, so this is just a bit faster than dialup. The kicker with 1xRTT is the the latency goes up to 400-800 ms.
If you are somewhere with "digital" coverage from Verizon, then you will probably be able to get decent 1xRTT coverage.
You might want to consider borrowing a friends verizon phone and see what type of signal it sees at your house. If you see a decent "1X" digital signal, then the 1xRTT will probably work ok.
If terms of what the service looks like while using it, it is a pretty typical ISP account. You get a DHCP public IP address. I have not noted any port filtering, but have not scanned looking (it does allow me to send email thru my corporate mail server on port 25). The adapter has a tendency to go into "dormant" mode, so you can see some lags if you need to wake up an ssh connection or similar.
In terms of end-user agreements, it is also a pretty standard client type service. No servers, no permanent connections, etc.
With these limitations, I am still very happy with the service. When they first came out with what was called "Express Network" (which was the first 1xRTT rollout), the connectivity was spotty, you would have to dial-in three or four times, and moving from cell to cell was a nightmare. For the last year or so, things have gotten a lot better.
In the end, you have to decide if $80/mo is worth it. For me, being able to reboot servers from the middle of the California desert, even at dial-up speeds, is a godsend. Plus my wife loves to IM with her friends at 70 MPH (of course not while driving).
My appologies for the inaccuracies in my description. I was not trying to imply that SIP used GET, but just a protocol like HTTP. Also, thank you for the information about the SIP spec requiring TCP. I knew TCP was in the spec, but thought it was rarely used. In fact, I think most providers firewall the TCP traffic anyway (we do that with our PBX here). I have a hard time imaging a SIP transaction that hits 1300 bytes anyway.
I think that SIP as a "peer to peer" protocol is the right description. Saying that peer to peer is also a server, while technically accurate, does not seem to be what ISPs prohibit. I have never seen a cable company complain about instant messaging as a "server application".
If you really want to have fun, configure SIP to run on port 53 (domain), at least until ISPs start transparent proxying domain queries so that they can act like Verisign and redirect mispellings to their own "contracted" search engines.
The real issue is that ISPs need to say what they are doing and state what they are charging for. If they have limitations, the limiations need to be non discriminatory and easy to understand. If a limitation is there to prevent abuse (things like blocking outbound port 25), then the ISP needs to state this and let users turn it off. The key is "disclosure" and "fairness". If ISPs disclose and have policies that are fair to competitors then it is hard to complain.
Slashdot Mirror
is very dependent on the application. In particular it depends on the mix of linear vs random operations and the mix of reads and writes.
For 100% read applications SSDs tend to be similar in performance to hard disks when reading linearly, and a lot faster than hard disks when reading randomly. This shows up in linear read speeds of 100 MB/sec for a typical Flash SSD which is "close" to a hard disk. For random 4K reads, Flash SSDs can stomp any hard disk. Most disks are in the 10,000 4K read IOPS range where 15K SAS drives are in the 250 range or 40x slower. So for applications that are 100% read SSDs can be as much as 40x faster, although the average is usually in the range of 15x to 20x.
When you start writing to Flash things get interesting. Flash is really designed for large, linear, aligned, writes. With most drives, you can get maximum write throughput only if you write exactly aligned with the drives internal erase blocks. Thus you can write exactly 2 megabytes on exact 2 megabyte drive boundaries and get 100% of the theoretical write throughput of the drive. Unfortunately, no application acts like this, so you are at the mercy of the file system and Flash controller to turn your smaller, probably random, and probably mis-aligned writes into what the drive can handle. The net impact of this is that good Flash SSDs have 4K random write IOPS in the 120s which is 1/2 the speed of a 15K SAS drive. I have measured Flash SSD with 4K write IOPS with values like 135, 120, 64, 43, 24, 13, 4.0, and 3.3.
This is why Flash SSD performance is so hard to judge. The random write performance can suck up the available "drive time" and dig a system deep into dirty buffer flushing. We talked with one Dell laptop user that described their system becoming "unusable" while an Outlook indexing operations was randomly updating a big file. Unusable in this case was 2+ minutes for to bring up task manager.
These random writes also have a real impact on the wear of the drive. Every time you seek a write, you basically chew up a write/erase cycle, even if the write is only 4K long. If you look at a drive that claims 50 GB/day for 10 years, this is 50 GB of linear writes on exact erase block boundaries. If you write 4K randomly, the 50 GB really means 25,000 4K writes or 100 Megabytes of random writes.
The solution to this is to not write randomly to the drive. There are file systems designed for Flash that address these issues. These are typically called "Log File Systems". Unfortunately, there is no generally available file system really designed for performance. In Linux the LogFS options are really tuned for small memory small storage systems and for hardware where the flash chips are directly accessible. They do help drive wear a lot, but they are just not tuned for Gigabytes of space or database crunching performance.
Another solution is my companies product called MFT (Managed Flash Technology) which is a software block mapping layer that runs on the host. It gives you the random write performance benefits and wear benefits of a LogFS while allowing you to use whatever file system you wish. MFT was developed on 2.6 Linux and has been ported to Windows. With MFT, the same drives that do 25 4K random write IOPS usually measure over 10,000. The linear speed of the drive is still equal to a hard disk, but the random speed is now closer to symmetric with reads and writes. Thus jobs like updating databases can literally run 20x faster than the fastest hard disks.
In the end, Flash SSDs will find specific markets initially. I can say with certainty that they won't get used for off-line backups or storing/edit large quantities of HD video. But give them databases or file systems with lots of small files, and they can really smoke a hard drive.
Most current Flash SSDs have a 2 Megabyte erase block. Take a 32 GB MLC drive with 10,000 cycles and you wear out in 160 million random write operations. At 20/sec this is 92 days. At 200/sec it is 9 days.
Good enough for laptops. Not good enough for servers.
On June 15th, Mtron will start shipping the 1000 series MLC drives. Put these in an array with the right software and you end up with price/GB parity with 36GB 15K 2.5" SAS drives and about 12x the random IO performance.
HDD Array:
8 Seagate Savvio 2.5" HDDs: $350ea $2,800
configured raid-10
1 SAS raid controller $600
Total cost for 144 GB $3,400 or $23.61/GB
SSD Array:
6 Mtron 1025-32 2.5" SSDs: $290ea $1,740
configured raid-5
1 SATA raid controller $250
MFT Software License $1,250
Total Cost for 144 GB $3,240 or $22.50/GB
HDD Performance:
4K and 8K read IOPS: 250/2000 (single-threaded/multi-threaded)
4K and 8K write IOPS: 1200
SSD Performance:
4K read IOPS: 8000/48000 (single-threaded/multi-threaded)
8K read IOPS: 6000/36000 (single-threaded/multi-threaded)
4K write IOPS: 40000
8K write IOPS: 22000
These performance numbers are with the MFT driver in place. Without MFT, the 4K random write performance is about 140 IOPS (>250x slower).
Endurance for these SSDs in this configuration is good enough to overwrite the entire array with random data three times a day (500GB of random updates/day) for about five years.
These drives make a wicked mail server (EasyCo just moved one of it's mail servers mirrored to MLC flash and the difference is amazing).
Sorry for the blatant advert, but SSDs are here now.
Doug Dumitru
EasyCo LLC
http://managedflash.com/
+1 610 237-2000 x2
we use dynatron A48G for AM2. Not quiet, but if the OS can fan control it stays reasonable.
I have build a couple of these for colos, but they are low cost, quiet, and low power, so you might be interested.
Start with a Supermicro 1U 510. This is 11" deep with an 80+ efficiency 200W power supply. It will house most uATX motherboards, although you will have to nibble out the back panel to get non supermicro boards to fit.
Then add a low cost AMD socket-AM2 motherboard.
Put a low power 45W dual core like a BE-2400 or one of the newer series. You will need a 1U blower for this.
Put one 3.5" drive (or up to 4 2.5" drives).
With a Gigabyte board and 1 2.5" HDD + 1 2.5" SSD, my system draws 37 watts in use as measured from the power line. It is not 100% quiet, but is the quietest of any 1U I have worked with.
Its what happens when techies run a company rather than marketeers.
My apologies for a long post. There will be some adverts embedded, but I will try to keep things informative.
The reason that Flash SSDs act "wierd" in benchmarks is that they have asymmetric performance patterns when reading and writing. Particularly with random operations, this asymmetry is huge. Here are a couple of example "drives":
* Mtron 7000 series: >14,000 4K random reads. ~130 4K random writes.
* SanDisk 5000 series: ~7,000 4K random reads. 13 4K random writes.
* Cheap CF card or USB stick: ~2,500 4K random reads. 3.3 4K random writes.
This is a 100:1 performance deficit when doing random writes versus the random reads. This has some really weird impacts on system performance. For example, if you run Outlook and tell it to "index" your system, it will build a 1-4 GB index file in-place with 100% random writes. If you do this on a hard disk, the job takes a long time and drags down your laptop, but the operation is still pretty smooth. Do the same think on an SSD and the system slugs to molasses. One of our customers described it as "totally unusable" with 2+ minutes to bring up task manager. What happens is that the fast reads allow the application to dirty write buffer faster and this then swamps system RAM, you get a 100+ deep write queue (at 13/sec), and you want to throw the machine off of a bridge.
This fix as some have described it is not some magic new controller glue or putting the flash closer to the CPU. It is organizing the write patterns to more closely match what the Flash chips are good at. Numerous embedded file systems like JFFS do this, but they are really designed for very small devices and are more concerned with wear and lifespan issue than performance.
Now here comes the advert (flames welcome). A little over 2 years ago, I wrote a "block translation" layer for use with Flash storage devices. It is somewhat similar to a LogFS, but it is not really a file system and it does not play be all of the rules of a LogFS. It does however remap blocks and linearize writes. Thus it plays well with Flash. It also appears to be an "invention", and thus my patent lawyer is well paid.
The working name of the driver layer itself is "Fast Block Device" (fbd) and the marketing name is "Manged Flash Technology". And what this does is to transparently map one block device into another view. You can then put whatever file system you want into the mix.
In terms of performance, it is all about bandwidth. Build a little raid-5 array with 4 Mtron drives and you will get over 200 MB/sec of sustained write throughput. With MFT in place, this directly translates into 50,000 4K random writes/sec. Even better, you tend to end up with something that is much closer to symmetric in terms of random read/write performance.
MFT is production on Linux (it has actually been shipping since last summer) and is in Beta test on Windows. It works with single drives as well as small to medium sized arrays. It does work with large arrays, but the controllers don't tend to keep up with the drives, so large arrays are useful for capacity but don't really help performance a lot. Once you get to 50,000 IOPS it is hard for the controllers to go much faster.
Consumer testing with MFT tends to produce some laughable results. We ran PCMark05's disk test on it and produced numbers in the 250K range. This was with a single Mtron 3025. Our code is fast, but we fooled the benchmark in this case.
There are several white papers on MFT posted in the news link of our website:
http://managedflash.com/
My apologies for the advert, but I see a lot of talk about SSDs without actually knowing what is going on inside.
I am happy to answer any questions on-line of off.
Doug Dumitru
EasyCo LLC
610 237-2000 x43
http://easyco.com/
http://managedflash.com/
http://mtron.easyco.com/
Swap to Flash is "stupid", but the stupidest in 55 years. Surely you can come up with something worse.
Anyways, swap of any kind only make sense if adding memory is inconvenient. It all depends on whether system memory is even available at the size you need.
Right now, RAM prices are so low that flash as swap does not make much sense.
Flash devices are fast at small reads but very slow at small random writes. Swap is usually a 50-50 mix of reads and writes with no pattern to be seen. Thus if you swap to Flash, at least traditional Flash, then you will be very unhappy.
...
Your comment about bandwidth needing to go to a spindle seems strange as well. I have a 4 drive raid-5 flash array here that just tested at >400 MB/sec on reads and >150 MB/sec on random writes
http://managedflash.com/news/papers/07-12-01_mtron-benchmarks.pdf
I don't know of many "spindles" that can keep up with that. Plus total power draw is 12 watts operating and 2 watts idle (actually the raid controller draws another 15, sorry).
If you want to see how Flash "can" be used for swap, see:
http://managedflash.com/
Even compact flash cards are effective for swap yielding about 2000 4K random read/write IOPS.
ps: sorry for the advert. at least it was short.
Random write performance to bare drives is usually quite bad. Most "reputable" vendors do publish random write figures. SanDisk quotes 13 IOPS in the spec sheets. Mtron quotes 120 IOPS. I have not seen quotes from Samsung, but have tested their old drives at 27 IOPS. I even tested one drive at 3.3 write IOPS.
On the other hand, random writes issues are "fixable". My company just published tests for various Raid-5 Flash SSDs setups. For 4 drives testing with 10 threads on Linux 2.6.22 using our MFT "driver", we get:
4K random reads 39,689 IOPS 155 MB/sec
4K random writes 29,618 IOPS 115 MB/sec
These are real numbers and the application does see the performance improvement.
For full details on drive performance see:
http://managedflash.com/news/papers/index.htm
1 megabit of 95% percentile bandwidth is about 100G/month. This assumes you are "using about 30%" of the peaks on average. You might be doing a little better. At 2T, this implies that you are 95% peaking at 20megabits. The dirt-cheapest you can buy a megabit for is the low teens, so this would be $250/mo in raw bandwidth costs assuming you are buying a lot more than 20. If you bought this directly from a tier-1, 20 megabits would be about $70-$90/mbit depending on where you bought it.
The bottom line is that I do believe that you are running 2T for $120. You are probably renting a basic dedicated server from ServerBeach or similar and they bundle in a lot of BW with the deal. They are playing the averages. If you are actually using 2T, then they are losing money on you. This is why these companies won't sell you 4T for $240. They also won't sell you 2T with your server and often restrict porn. They really want the averages to run 100G/system.
DailyKos has >1M visitors/day. This is not page hits, but unique people. Do the math. 1M * 5 pages * 100K/page (they are big pages) = 500G/day or 15T/mo. This is ~150mbits bursing to ~400mbits, so we are talking a dedicated gig-e port. 200mbits of level-3 from colocation.com costs $50/mbit, so that is $10K for BW. It appears that Kos is not getting that bad of a deal after all (plus I suspect the $7K including things like racks and power).
Your getting 2T for $120 is akin to the Lieberman campaign. You get what you pay for.
I you use enough bandwidth, you will get a $7K bill. What the author was saying is that he has a $7K monthly bill every month and expects it.
If you think you can cheap out and push terabytes for nothing, you will quickly find out that free lunches are hard to find anymore.
Now exactly how youtube.com stays in business, I don't get.
If you don't care how warm it gets, you can build a server-class laptop pretty easily. Go to alienware.com and configure an:
* Athlon 64 X2 4800
* 2G of ram
* 2x 100G SATA 7200 drives
Sounds like a server to me.
I am happy to agree with you, but I suspect that I will draw the line at a quite different point than you might.
If you want to argue that a candidate cannot send you an email once a day (or once an hour) then I think you are wrong. If you want to argue that a candidate cannot send you 1000 emails per hour shutting down your inbox, then I think you are right. You really do have to go to extremes to find cases where restricting political speech is justified and constitutional.
If the speech itself makes your inbox useless because it crashed your mail server it is probably over the line. I would argue that it is mostly over the line because it would prevent other political speech from reaching you.
Regardless, a candidate making phone calls, to your cell phone, that is unlisted and on the do-not-call list, a couple of times before an election, is protected, even though it will bug the hell out of you. The alternative of restricting this is a cure that is worse than the disease. As email goes, this means that you have to put up with a dozen unsolicited emails from a candidate or advocacy group and it will bug you, but it is not reasonable (or smart) to ask for a law to prevent it. If you don't want to read the email, don't. Engage the sender. Campaign against them. But don't expect a law to protect you from being annoyed.
Remember, there is a constitutional right to free speech. There is no constitutional right to not be offended (or annoyed).
There is the obvious issue of "cost to deliver", but that does not change the speech. If you rent a mailbox at the local post office, junk mail and can does cost you real money. This is probably a lot more money than your email box.
If someone sends you an email, they can only send it to you because you have setup your mail server to receive it. They did not break into your computer system to deliver the mail to you. You setup a server that is designed to accept exactly what the spammer is sending to you. You might "prefer" that he not send this stuff to you, but you have not communicated this preference to the sender in any way, let along entered into a contract with him so that he agrees to pay you.
If your inbox is your property, then set it up so that it no longer accepts email from anyone. It is your property and you are prefectly within your rights to refuse to receive mail from anyone not on your white-list. If you don't setup mail this way, then this is your choice. If you want people to pay you for the priviledge of getting access to your email inbox, then setup your email inbox to send back a contract that the sender can sign. Just because you "want" something to work in a particular way does not mean that the other party agrees with you.
So you can use your inbox how you want to. You could have restricted access to it, but have chosen not to, or are just too lazy to want to go to the trouble. Your solution is to have the government jail some politician because he is somehow violating your property rights, even though you have done nothing to protect them. Your inbox is an open, green, field with no fence, no "do not trespass" signs. Nothing to indicate that everyone is not welcome. You have elevated your right to not be annoyed and minor incremental costs of running an inbox (of which 99.9% are the cost of your time, ie. your annoyance, and not the actual cost of the inbox) over the constitutionally protected right of free political speech.
If this were an UCE (Unsolicited Commercial Email) and not your expanded definiation that include political, religious, and other non-commercial email in the broad term of SPAM, then you might have a case. The constitutional protections on commercial speech are much lower. But this is political speech, so you are just plain wrong.
I get really amazed at the people that think that SPAM is equal to mass murder.
SPAM is many things to many people. To some, it is an email message, commercial in nature, and unsolicited. To some it is any unsolicited email. To others, it has to be sent in bulk (whatever that means).
Regardless, if the email itself is political in nature, it is covered by free speech. If you don't want to read it don't. If you want to filter it out fine. If you want to only get email from your frields, blacklist everyone else. Don't somehow claim that political emails are not free speech. They ABSOLUTELY are free speech. For you to argue otherwise is simply amazing.
And yes, a politician can come to your front door (during reasonable hours), ring the doorbell, and ask to talk to you even if you have a "post no bills" and "no solicitiing" sign prohibiting this. This is free speech. If you are registered to vote, the political parties and candidates can get your name and address and can contact you in whatever way they can figure out that does not cross the line into harassment and trespassing. You cannot opt out of this. If you want to opt out, then you cannot register to vote. Sorry. This is how the system works (at least in California). And this is a good thing because voting and political speech is far far more important than spam.
When I lived in Georgia a couple of decades ago, we would occasionally see the Klan going door to door and standing on the street corner. I did not agree with it, but it was free speech. It was disruptive. You could even argue that it was dangerous and evil. It was still free speech and as such protected.
Remember guys that getting an unsolicited email is not mass murder. In political terms, this is advocacy. You can complain that using email is "doing it on the cheap", but then again, giving lesser financed voices a stage is probably a good thing. Your power against this lies with shaping public opinion, voting against a bad policy, or advocating against the positions that you disagree with. To say that political speech does not belong in email is just unamerican (and I use that phrase intentionally).
Your logic is all wrong. You have setup a mailbox that accepts, from the public, inbound communications. This is no different than having a mailbox with a street address. By setting this up, you are accepting speech from the general public, including from those that are anonymous. You did this by your choice. Now you are saying that the government, using its ability to commit violence on its citizenship, should pass and enforce a law that restricts what someone can say to you, and moreso, what someone can say as political speach.
Receiving an email is not "taking over your living room". That is the stupidest analogy I have ever seen. Getting junk mail in your post office mailbox does not require you to listen. If you are renting a mailbox at the local post office, the anoyance factor of junk mail is about the same as filling your email inbox (and the actual hard cost is probably more). Free speech means that people are free to talk. It means that they are free to talk to you. It does not mean you have to listen, but it does mean that you have to put up with the annoyance of them "trying" to talk to you. Sorry, but you do not have a constitutional right to not be annoyed.
And I am very happy that you are not an elected official or judge with the responsibility of making laws or interpreting laws or the constitution. Free speech is much more important than your inbox. Sorry if you disagree with that.
But you are wrong. Email is absolutely speech. Your argument is that receiving an email is somehow trespassing. Unfortunately, you invited this trespass (or at least you did not take any action to prevent it). It was you and your ISP that made the decision to receive un-authenticated email from the public internet. If you don't like the email, filter it, don't read it, or just delete it. You are not required to read it anymore than you are required to read a billboard on a public sidewalk.
The real issue here is what is more important. Free, political, speech, or the convenience of your inbox being less cluttered. If you think your inbox is more important, then you just don't get it. Some things are truely more important than convenience and making political speech, even if it is annoying, illegal is a very bad idea and the definition of the "solution is worse than the problem".
The CanSpam law was already on somewhat shakey legal ground. In the US we have the 1st ammendment, you know, the one about free speech. Regardless of how much you might dislike it, spam is a form of speech. Usually, it is a form of commercial speech.
The interpretation of the constitution deals with speech issues differently depending on the type of speech involved. There is very little protection for fraudulent speech, so this can be made a crime and you can be put into jail. The same goes for yelling fire in a theater.
Commercial speech is somewhere in the middle and the government can regulate it, but only at the edges. In general, current constitutional interpretations allow for almost all commercial speech.
Political speech is at the other extreme. Whereas fraudulent speech can be outlawed and commercial speech can be limited, political speech is "the most protected" type of speech. Political speech might be annoying, might make you angry, and might even be racist or offend you. Regardless, if it is political speech, then it is allowed, pretty much without restriction. The constitution guarantees you the right to free speech. It does not guarantee you any right to not be offended or annoyed.
Political speech is also not limited to a candidate running for office. Groups often make unsolicited calls, ignoring the do-not-call lists, on behalf of an issue or cause.
Before you complain about this and suggest that "there needs to be a law", think very carefully. First, I doubt that a law would do it. The free speech protection is in the constitution so a law might not be valid anyway. Regardless, is your comfort at the dinner table or at your PC more important than the constitution. You always have the right to not answer the phone, email back to jerks that pollute your inbox, call them to complain, waste their time (while being careful to only express your opinion and not harrass). It would be a very bad idea to want the government to "pass a law" allowing the government to "thru the use of force" put someone in jail because of what they say, especially when the topic of the speech is political.
The bottom line here is that there are other ways to achieve balance. If a campaign uses email and does not manage their lists and remove requests well, don't vote for them. This is not a no vote because of disagreement with the candidates positions, it is a no vote because of a candidates lack of organization and poor adminstration of their data. If a candidate cannot properly handle an email list, do you realy expect them to be able to run a government office.
The other point here is that not voting for a candidate just because they sent you an email or called you with an automated message might be really stupid on your part. If you are too lazy (as many US voters are) to vote anyway (let along actually check out issues and learn about your local candidates), then perhaps you deserve to be interrupted at dinner. I find that the best way to turn a candidates machine off is to say "I have already voted by mail". Gets them every time.
Level-3 and Cogent are not all that differnet in size. They both cover the US from coast to coast. They both go to Europe. Level-3 goes to Hawaii, but that is the only real footprint difference.
Cogent has fewer US POPs, but still hits all four corners. Cogent claims 80G backbones. Level-3 claims 110G backbones. Both networks have real 24x7 operations centers and skilled operators.
You cannot deny that there is a differnce in size between the two companies, but it is pretty easy to describe both as "Tier-1" networks.
If you read the Level-3 peering policy statement:
http://www.level3.com/1511.html
(which is where the Level-3 bw numbers come from, so they might be a little old), it would look on the surface that peering with Cogent is perfectly reasonable.
The complaint of Level-3 is twofold. The first one is, "we are bigger, so you should pay us". It is true that L3 is bigger, but whether they can throw that around and make Cogent pay is not 100% clear. When you buy from Level-3, you "expect" to be able to talk to Cogent. I am a Level-3 customer and L3 did not imform me that they were cutting these lines off. I was pissed. I knew that L3 could legally do what they wanted, but I thought it inappropriate for them to do it without giving their customers notice. Level-3 chopped off 45 million IP addresses when they de-peered Cogent. Even if L3 thought it was Cogents "fault", it was still L3 that threw the switch.
One other issue of this is that some major suppliers buy from L3 and don't multi-home. Thus, RoadRunner cable modems lost contact with all Cogent sites. This left RR answering the phones from customers. I am sure it cost RR a lot of hassle, and probably some real money. I cannot imaging that RR was not threatening L3 with "we are going to multi-home and you will lose 60%+ of our traffic if you don't get this working now". L3 had a lot to lose.
This is the real problem with Level-3 taking action. Whether Level-3 is right or wrong, it's customers will still blame L3. L3 did enter into peering with Cogent. L3 did initiate the termination of the peering arrangement. In doing so, L3 did cutoff a large chunk of the internet to it's own customers. Image if Level-3, on it's home page, prominently stated that you were not buying connection to the "whole" internet. No one would buy service from them. Level-3 gets customers because, on the whole, we trust Level-3 to work hard to keep us connected. This trust is what L3 is selling. The de-peering has damaged this trust and will, in the end, cost L3 a lot more money than they were trying to extract from Cogent.
The only real, technical issue between L3 and Cogent is traffic symmetry. Cogent sells to a lot of hosters. L3 sells to a lot of end-users (cable modems, dial-up, etc.). Thus more bytes move from Cogent to L3 than from L3 to Cogent. If the two networks only touched at one point, this would still be fair. After all, it takes a L3 customers to request a Cogent porn page. But the peering between L3 and cogent happens at multiple points. Thus, you get the "hot potato" case where traffic from Cogent dumps to L3 early (and vice versa, traffic from L3 dumps to Cogent early). This means more Cogent bytes travel on L3's network than on Cogents network. This is presumably what L3 wants Cogent to pay for. While this may be a "real" traffic issue, I suspect L3 wants more from Cogent than is actually reasonable. What would be interesting would be for L3 to advertise its routes to Cogent differently at each peering point. Thus Cogent would carry it's own outbound traffic over more of it's own network and the "hot potato" effect would not happen. Perhaps BGP does not work that way, but it should be able to. I suspect L3 was more interested in "monetizing" it's relationship than in making it "fair".
I also just noticed that L3 is starting to compete with Cogent on price. It used to be that the $3000/mo for a 100mbit
You are right:
For every republican you can find that's corrupt I can find a democrat...
but then again, it took the dems 40 years to screw things up, and the repubs have done it in 10.
The problem is not who is dishonest, but who is dishonest and efficient. Repubs and dems are both dishonest (or have dishonest members among them), but the repubs are head and sholders above dems at being organized, zelous, dogmatic, controlling, self-rightous, etc.
And in talking about Clinton, I will take the midget of the cabinet (Robert Reich) over each and every official on the W team.
I don't know of an existing tool for this, but you want to measure "known traffic" across the route and report when it degrades.
Setup an "application ping server" on the far end. This will be a C program that posts a UDP listen on a port in the RTP range. When it gets an inbound packet, it returns exactly that same packet to whom sent it to them. This needs to be in C (or similar) because the latency needs to be very low. It should also run on a very low utilization server.
On the near end, write a similar program that sends a UDP packet to the far end and measures round-trip delay. If you want to really act like a VOIP call, send 200 byte packets at 50/sec (or whatever the SIP/RTP payload is for your particular codec).
The question is then how to analyze the data. I would probably have the program spit out peak/avg "numbers" for latency and packet loss into a text file, perhaps at 1/minute. You could then have MRTG (or your favorite grapher) read those and draw you pretty stuff. Once you see the types of degredation that actually matters, you can program in your thresholds to page/email/whatever you.
Things to remember when setting this up. Make sure the traffic is handled by the routers exactly like your VOIP packets. Run on the same UDP ports with the same QOS tags. In other terms, make your test traffic indistinguishable from the actual voice traffic (at least from a router's/switch's voicepoint).
Of course, someone probably has a package that does this. If not, build it and open a sourceforge project.
If I read this story correctly, the chain of events appears to "maybe" be a bit more interesting (but I am speculating on some of this).
...
...
...
* Borders was apparantly not paying sales tax (speculation)
* They decided to pay the sales tax
* This might have been encourages by the California Tax
Amnesty program (again, speculation)
* ie, no penalties, but still interest
* Ahhhnolds plan to help balance the budget
* Borders then sues the state to get the money back
* They lost
* Borders then appealed
* They lost again (this is the current case)
In thinking of how borders runs their business, the line
If it looks like a duck
seems to apply.
If you don't like taxes, remember that for every tax that is avoided, there has to be another tax to make up for it.
Of course there is always the line
We should tax all foreigners living abroad.
I have a Verizon account. I have actually had it for a couple of years and it has improved markedly over time.
/. article on the "Stompbox" Verizon to WiFi mobile hot-stop.
b roadband/index.jsp
Verizon has two classes of available service. You can use a single PC-CARD adapter that will auto-switch depending on which service is available to you. There is even another
If you are lucky enough to be in a Verizon EVDO "Broadband" access area, you can expect 2 Megabits "peak" downstream speed. Not sure what the peak upstream speed is. Regardless, if I have reasonable single strength, I get about 500kbit down and 80kbit up. I think it is fair to call this "DSL speed". Even though the speed is good, the latency is still a bit high at about 200-250ms.
At this point, the "Broadband" access areas currently cover about 30 cities. If you are really "out in the woods", I would be very surprised if you are in one of these areas, but you can check. This is a list of the broadband access cities:
http://www.verizonwireless.com/b2c/mobileoptions/
If you are not in one of the "Broadband" areas with EVDO coverage, you can still use the 1xRTT service throughout most of the rest of the country. 1xRTT gives you theoretical 144kbit down. I tend to get about 60-80 kbit down and 20kbit up, so this is just a bit faster than dialup. The kicker with 1xRTT is the the latency goes up to 400-800 ms.
If you are somewhere with "digital" coverage from Verizon, then you will probably be able to get decent 1xRTT coverage.
You might want to consider borrowing a friends verizon phone and see what type of signal it sees at your house. If you see a decent "1X" digital signal, then the 1xRTT will probably work ok.
If terms of what the service looks like while using it, it is a pretty typical ISP account. You get a DHCP public IP address. I have not noted any port filtering, but have not scanned looking (it does allow me to send email thru my corporate mail server on port 25). The adapter has a tendency to go into "dormant" mode, so you can see some lags if you need to wake up an ssh connection or similar.
In terms of end-user agreements, it is also a pretty standard client type service. No servers, no permanent connections, etc.
With these limitations, I am still very happy with the service. When they first came out with what was called "Express Network" (which was the first 1xRTT rollout), the connectivity was spotty, you would have to dial-in three or four times, and moving from cell to cell was a nightmare. For the last year or so, things have gotten a lot better.
In the end, you have to decide if $80/mo is worth it. For me, being able to reboot servers from the middle of the California desert, even at dial-up speeds, is a godsend. Plus my wife loves to IM with her friends at 70 MPH (of course not while driving).
My appologies for the inaccuracies in my description. I was not trying to imply that SIP used GET, but just a protocol like HTTP. Also, thank you for the information about the SIP spec requiring TCP. I knew TCP was in the spec, but thought it was rarely used. In fact, I think most providers firewall the TCP traffic anyway (we do that with our PBX here). I have a hard time imaging a SIP transaction that hits 1300 bytes anyway.
I think that SIP as a "peer to peer" protocol is the right description. Saying that peer to peer is also a server, while technically accurate, does not seem to be what ISPs prohibit. I have never seen a cable company complain about instant messaging as a "server application".
If you really want to have fun, configure SIP to run on port 53 (domain), at least until ISPs start transparent proxying domain queries so that they can act like Verisign and redirect mispellings to their own "contracted" search engines.
The real issue is that ISPs need to say what they are doing and state what they are charging for. If they have limitations, the limiations need to be non discriminatory and easy to understand. If a limitation is there to prevent abuse (things like blocking outbound port 25), then the ISP needs to state this and let users turn it off. The key is "disclosure" and "fairness". If ISPs disclose and have policies that are fair to competitors then it is hard to complain.