On Fri, 11 Jan 2002 10:39, Anders Gjære wrote: > I have a router running BGP / Zebra, and it seems like the maximum > throughput is 25Mbit/s > > BGP and Zebra using 100% cpu together, and alternating on witch using > most.
Presumably you mean that the bgpd and the zebra process use 100% CPU. When you say that the maximum throughput is 25Mb/s, is that for routing or for BGP protocol? Do the zebra and bgpd processes use any significant amount of CPU time when the system is not routing any data? If there is no data being routed the zebra and bgpd processes will have to do the same amount of work. So if stopping the routing decreases the apparent CPU usage of them then they are not really using CPU time. When the kernel receives an interrupt for a network packet it will route it, and put the time taken to do so against it's count of the time used by the application that was running at the time of the interrupt. For example, I was once running a rather slow workstation on a 10base2 network that had a lot of Netware traffic which involved lots of small packets. After accidentally putting the network interface into promiscuous mode the most trivial processes would appear to take 100% CPU time (of course the machine in question was a 486 with an ISA network card). With a modern machine such as your P2-233, it should be able to route 100Mb/s easily if you have good network cards and firewall rules that aren't too complex. I always use Tulip cards when given a choice, as they seem to be able to do a lot of work without using much CPU power. But I'm not aware of any good survey of cards in this regard. -- http://www.coker.com.au/bonnie++/ Bonnie++ hard drive benchmark http://www.coker.com.au/postal/ Postal SMTP/POP benchmark http://www.coker.com.au/projects.html Projects I am working on http://www.coker.com.au/~russell/ My home page
