Xen Project Q&A Forum: First Line Help for Simple Questions
This is your chance to ask questions and provide answers about basic use of the Xen Project software. For debugging problems and for more complex issues, consider using the xen-users mailing list instead. You can find information about xen-users under "HELP | Mailing Lists" in the navigation bar above.
Additional info! Two centos systmes, One of them is Xen 4.2 host(Dom0). Checked with varying buffer sizes in /pro/sys/net/ipv4, but same result. Latency is 0.2 ms. I have set up bridge for eth0. Is linux bridge and Xen bridge (backend) is the same, I am little confused a bit here.
After this I want to test throughput of VMs sitting on Xen 4.2 with varying amount of MTU.
Actually I am doing my thesis on virtualization and I chose Xen
Thanks guys ! Any help much appreciated....
Accepted AnswerRussell PavlicekOffline0Well, I have little experience mucking with MTU settings, but I did find a little info on the subject.
There is an interesting (but, sadly, undated) research paper out of Japan which might be helpful:
An HP support article of unknown vintage says setting MTU around 1500 is optimal:
If you want to use Jumbo Frames, it appears that MTU=9000 is best. See:
Someone else was trying to adjust network performance using the MTU only to find that enabling GSO changes things:
There's also an old (2006) paper which discusses Xen MTU settings, but I don't know if it is still accurate:
If nothing here satisfies, then I'd suggest querying the xen-users mailing list. You will find a number of Xen developers who hang out there. They might be able to give you additional insight.