>
> The server runs on a Core2Quad 9300, with 8GB RAM (max motherboard can
> take, 1U chassis) on an Intel motherboard with a 1TB SATA HDD.
>
> dom0 is set to 512MB limit with a few small XEM VM's running:
>
>
> root@zaxen01:[~]$ xm list
> Name ID Mem(MiB) VCPUs State
> Time(s)
> Domain-0 0 512 4 r-----
> 96.5
> actionco.vm 3 1519 1 -b----
> 14.8
> byracers.vm 4 511 1 -b----
> 85.7
> ns1 5 511 1 -b----
> 22.3
> picturestravel 6 255 1 -b---- 13.3
> rafttheworld 7 255 1 -b----
> 11.3
> zafepres.vm 8 511 1 -b----
> 19.0
>
...
What are the actual symptoms you are seeing?
In general I found that tuning the disk scheduler and also the Xen guest
scheduler to be helpful:
http://wiki.xensource.com/xenwiki/CreditScheduler
http://www.cyberciti.biz/faq/linux-change-io-scheduler-for-harddisk/
Also, I always recommend building Xen servers to use SAS drives rather than
SATA because SATA are half duplex while SAS is full duplex, meaning under
higher or more random IO you will better throughput. In my experience I
see almost double the performance when using SAS over SATA, but our
environments are IO heavy and may not reflect the realities of your
environment.
I would also suggest, running disk IO stats in the VMs simultaneously while
running iostat or vmstat in Dom0 to get a good read for where bottlenecks
really are. I actually prefer to use the simple postmark utility as it is
relatively simple and avoids disk caching issues which skew your results.
_______________________________________________
CentOS mailing list
CentOS@xxxxxxxxxx
http://lists.centos.org/mailman/listinfo/centos
