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/"
"According to the Internet2 LSR contest rule #5A, IPv4 TCP single stream"
vodka, straight up, thank you!
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
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.
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.