Slashdot Mirror
NetBSD Sets Internet2 Land Speed World Record
Daniel de Kok writes "Researchers of the Swedish University Network
(SUNET) have beaten the Internet2 Land Speed Record using two Dell 2650 machines with single 2GHz CPUs running NetBSD 2.0 Beta. SUNET has transferred around
840 GigaBytes of data in less than 30 minutes, using a single IPv4 TCP stream, between a host at the Luleå
University of Technology and a host connected to a Sprint PoP in San Jose, CA, USA. The
achieved speed was 69.073 Petabit-meters/second. According to the research team, NetBSD was chosen 'due to the scalability of the TCP code.'"
"More information about this record including the NetBSD configuration can be found at:
http://proj.sunet.se/LSR2/
The website of the Internet2 Land Speed Record (I2-LSR) competition is located at:
http://lsr.internet2.edu/"
SUNET Internet2 Land Speed Record: 69.073 Pbmps
Now give me my carma..
"According to the Internet2 LSR contest rule #5A, IPv4 TCP single stream"
vodka, straight up, thank you!
It's a measure of transport capacity.
"Analogous to man-miles/year considered by airlines"
And like the anonymous comment above mine, also analogous to gigabit-miles/hour.
A search of bit-meters gives you some references, however helpful they actually may be.
a) Because UDP isn't reliable. In other words, if you'd send byte by byte via UDP, there's no guarantee that they arrive in the same order, or arrive at all. You'd need to make it reliable by implementing your own layer on top of it. b) because TCP is what the majority of traffic on the net is, and using it for a benchmark is more realistic
Assume I was drunk when I posted this.
Perhaps because they wanted the data to arrive reliably?
UDP just sends off the data without caring whether it actually arrives intact at the other end, you know. TCP, on the other hand, actually gives delivery guarantees...
Re: your sig...
;)
To provide more relevacne for the band you might want to use something like the following:
Googling up my brother's Acid Metal band, Ahymsa
Google places more weight on the text that's actually inside the link
Straight from the site:
..assuming you were speaking of mbits, too. :P
838860800000 bytes in 1588 real seconds = 4226 Mbit/sec
Use Minidisc? Join the Minidisc.org forums.
True. However, the various stacks have diversified a lot since the original BSD 4.4 stack. As a result, many of the TCP/IP stacks have different performance characteristics and features. AFAIK, the three main BSDs have kept their stacks in sync because they've been sharing code. A stack from NetBSD should be almost the same as a stack from FreeBSD.
Javascript + Nintendo DSi = DSiCade
I'm betting it's not a "land" speed record, seeing as how the data probably jumps through the air (satillite/microwave transmissions) at one or more points.
Nope. The vast majority of phone & data runs over fiber, without satellite or microwave. The latency on satellite is much worse, & microwave is more expensive. Fiber is the first choice.
(Not to mention the fact that being on, over, or under the surface of land or water means nothing to a data cable.)
Well, back when I worked for JDS Uniphase during the tech boom, there was a world of difference. Getting parts qualified for underwater cables was much harder. The cable owners don't want to have to send out a ship to pull a cable up off the ocean floor to fix it - it's very very expensive.
JDS had to guarrantee that they would make no changes in its production process without the approval of the customer, and JDS had to get similar guarrantees from its suppliers. Of course, JDS charged a lot more for undersea components, but reliability was much more important than cost.
And many customers would demand that the parts be made in North America - they wouldn't accept made in China or Taiwan.
Sigh. I miss working at JDS.
Well, this is the "Land Speed" record, so distance does matter to some degree. This makes it useful to compare against the "bandwidth of a station wagon" -- more of a comparison of amount and distance over time.
10b||~10b -- aah, what a question!
The Internet2 is a separate network running on IPv6. Currently it is being developed and tested between a veriety of universities, ballsy ISPs and a few buisnesses. Simply upgrading the current internet won't solve many of the problems. (like multicast) Supposedly once internet2 is doing really well, isps will slowly migrate until the old network is mosly gone.
Note, there are bridges between internet1 and internet2.
No, Internet2 is a network.
:)
Yep, thats what I said.
but I know for a fact that IPv4 runs over it too
Uhhh, That is also what I said, there are bridges between the two. IPv4 is encapsulated in ipv6 and every ipv4 address actually has a counterpart within the ipv6 address space.
Glad we agree
AFAIK, everything modern uses the BSD stack. Windows uses the BSD stack for crying out loud (a little known fact). As OS X is heavily based upon BSD, it too uses the BSD stack.
Not sure about linux, though. I wouldn't dobut that their stack is BSD-based (at least the parts that weren't stolen from SCO)
-- If you try to fail and succeed, which have you done? - Uli's moose
We just worked this out...
A Saturn Ion station wagon with the back seat folded down, full of LTO2 tapes, is 418 petabit m/s at 60 MPH, or about 6 times more bandwidth.
And about $600k worth of tapes.
You can work out the DVD bandwidth yourselves.
Latency sucks, though.
"Inband compression" refers to compressing the sent data stream. He's asking if the data being sent is compressed at the transport layer.
For example, a 28kbps modem can transmit data at 28kbps. But if the transport layer can perform compression, applications that send easily compressible data could see the link speed as being much higher. At 4 to 1 compression, the application would think it's a 112kbps link (when it's actually much slower).
The question matters because truly random data can't be compressed. But if the data they were using to test Internet2 was not random, a simple compression scheme in the transport layer would easily be able to bump up the apparent data speed - but because the gain would not be the same for all data, this would technically be cheating.
Windows does not use the BSD TCP/IP stack.
Yes, it does. It's gone through a few generations of development (having been a 16bit Win3x port that was bought off by Microsoft), but its origins are still BSD. Of course, there's been so much development on it at this point that it's difficult to recognize.
Javascript + Nintendo DSi = DSiCade
More precisely, it went from
San Jose CA to
Stockton CA to
Kansas City MO to
Fort Worth TX to
Pennsauken NJ to
Relay MD to
Chicago IL to
New York NY to
Manasquan NJ to
Tuckerton NJ to
London UK to
Brussels BE to
Amsterdam NL to
Hamburg DE to
Copenhagen DK to
Oslo NO to
Stockholm SE (where it changed carriers) to
Vasteras SE to
Gavle SE to
Luleå SE.
Or maybe it was the other direction; the site doesn't say clearly which way the transfer was.
http://alternatives.rzero.com/
Perhaps you should tell that to the guy who wrote the windows stack. He's posted before saying that it is not bsd.
Better double check that.
Javascript + Nintendo DSi = DSiCade
One of the insurmountable limitations of geosynchronous satellite communications is the nearly 45,000 mile trip the signal needs to take getting from point A to point C. It introduces a delay of almost a quarter second, and the signal attenuation over that distance limits how much data can be sent reliably. Surface-to-surface microwaves suffer from interference that reduces their transmission rate, and line-of-sight limitations. No, a strand of glass or copper hugging the crust of the planet is far faster than either, and would almost have to be used for any envelope-pushing stunts like this.
http://alternatives.rzero.com/
other poster is whacked.
internet2 is by no means IPv6 only. and i doubt they tunnel everything. they may have native IPv6 available everywhere but it would be too much of a pita to manage all of the tunnels for everybody who doesn't do IPv6 yet.
internet2 is nothing more than a private network with a minimum entry level and big pipes.
besides internet2 is old-news. all the cool places are working on the national lambdarail project. no routers, no teleco, dedicated lambdas on demand end to end. 10Gbit ethernet coast to coast.
Internet2 is a separate network in the same sense that Sprint and UUNET are separate networks. It's funded for academic use, and has a rigourously-enforced AUP which results in it not being used for commercial purposes at all. It is not an upgrade of the current Internet. The name "Internet2" does not signify "the replacement for the Internet"; if you're being charitible, it means "let's see how people use a multi-gigabit network if they don't have to pay for traffic". If you're being mean, it means "let's see if we can attract federal funding by sounding like we're exciting and important."
Many commercial networks (Level3 and UUNET spring immediately to mind) run commercial networks which are far closer to the bleeding edge than Internet2 is, in terms of the complexity of the routing system and the forwarding path. There are commercial operators who operate parallel 8xOC192 circuits which are routinely filled to near-congestion conditions 80% of the time (yes, that's an aggregate of 80Gbit/s between just two sites). The Internet is orders of magnitude more complex and advanced in terms of forwarding capacity than Internet2. There are commercial ISPs who sell production IPv6 services. There are more commercial ISPs who sell production IPv4 multicast services.
No ISPs will migrate to Internet2, since Internet2 is funded specifically for non-commercial traffic. There are no "bridges" between the private network known as "Internet2" and the Internet in the way that you imply; there are simply universities who are connected both to the private network called "Internet2" and to the Internet via commercial providers.
The private network known as "Internet2" is not an IPv6-only network. It does not feature a policy of shipping IPv4 traffic purely encapsulated within IPv6.
Hope this clears up a couple of things.
similar but not the same... :D
NetBSD, FreeBSD and OpenBSD
performs differently in terms of scalability etc..
as Mr. Felix von Leitner once demonstrated http://bulk.fefe.de/scalability/ (isnt up anymore maybe he's busy with new benchmarks)
Well, many of the technologies developped on this network will be transfered to the Internet some day.
In fact, you can see Internet 2 as a part of the internet. It's only that the BGP routing is done in a manner that it's only data between two academic nodes that pass through it.
Ex.:
1) college.edu slashdot.org -- Commercial internet
2) college.edu othercollege.edu -- Internet 2
Oh.. and we call Internet 2 that way in the US only. In Canada for example, it's CA*net4, but it's the same thing.
Take a look at readable tcp dump and you'll notice that it is just the ascii character set shifted continuously. Now if you NEVER need disk access then this could be usable (aka isp and router junction points) but once you hit disk you are bottlenecked. Even with U320 SCSI you can only hit 320 MB/s (~2.5Gbit/s) assuming linear reads at full cacity of your full array of disks.
Disk is limiting pretty much anything, such as playing raw 2K video (2048x1556) in real time (seconds is relatively easy but minutes is difficult). I could care less how fast your network speed as when 1 non-solid state device (ie. disk) is entered into the mix the network performance is notional compared to real performance.
"Survival of the fittest Max, and we've got the fucking gun!" - Pi
Yep, Windows did use BSD stack back in the days of NT 3.5 and 4.0, but the stack has since then been rewritten from scratch, Windows 2000 and XP does not contain BSD stack.
And Linux does not use BSD stack eighter. Linux kernel hackers have written their own stack too.
Of course you can build error correction and recovery into UDP. At which point you have reinvented TCP. Congratulations.
Finally! A year of moderation! Ready for 2019?
Actually, we did tests with Linux (both 2.4 and 2.6) and FreeBSD also, but with not as good results. Linux IP stack eats much more CPU (and memory!) than it should. Basic problem is the network buffer implementation (or the lack of!). This is true for both 2.4 and 2.6. A redesign is needed of the IP stack to make it perform better. FreeBSD have a lot of linear searches in their IP stack left, fixing that would most likely give the same result as for NetBSD. I may port over some of the NetBSD changes if I get some spare time. NetBSD had already fixed (most of) those problems, some of them long ago, therefore it was simple to just use it.
Helooo!
WAKE up!
Regents of UC Berkeley removed advertising clause in 1999. NetBSD, FreeBSD and OpenBSD took their time but there is practically no advertising clause in their source code licenses. If somebody really wanted to use BSD stack without that clause he/she might just ask authors to remove it.
Remember, the FSF has an ideological goal
that's really sucks - messing technology with ideology - disgusting..
True, the three BSD inherited the same TCP/IP stack. But the record is also the consequence of NetBSD's VM implementation, which has been completely rewrote from scratch.
NetBSD TCP/IP stack makes use of some NetBSD VM features which avoids copying data. NetBSD is able to send data using TCP with zero copies between kernel and userland. This is why it performs much better than other OSes.