Re: [Xen-devel] Low Network Throughput between HVM Guests

["Mr. Teo En Ming (Zhang Enming)" <space.time.universe@xxxxxxxxx>]

Hi Pasi,

I have added 'type=ioemu' and 'model=e1000' to vif in my HVM configuration files. Now the Rocks HPC Cluster 5.1 x86-64 Frontend node detected two Intel 82540 Gigabit Ethernet controllers. Hopefully this will speed up my compute node installation with the upgraded gigabit speed.

-- 
Mr. Teo En Ming (Zhang Enming) Dip(Mechatronics) BEng(Hons)(Mechanical Engineering)
Alma Maters:
(1) Singapore Polytechnic
(2) National University of Singapore
My Primary Blog: http://teo-en-ming-aka-zhang-enming.blogspot.com
My Secondary Blog: http://enmingteo.wordpress.com
My Youtube videos: http://www.youtube.com/user/enmingteo
Email: space.time.universe@xxxxxxxxx
Mobile Phone (Starhub Prepaid): +65-8369-2618
Street: Bedok Reservoir Road
Country: Singapore

On Sat, Nov 7, 2009 at 9:15 PM, Pasi Kärkkäinen <pasik@xxxxxx> wrote:

On Sat, Nov 07, 2009 at 07:23:48PM +0800, Mr. Teo En Ming (Zhang Enming) wrote:
>    Hi Pasi,
>
>    I am not sure whether I can install PV drivers in Rocks HPC Cluster
>    software 5.1 x86-64 (based on CentOS) HVM virtual machines. When I tried
>    to install Rocks 5.1 x86-64 as a PV guest, it complains "invalid kernel"
>    and something about the ELF kernel image. So I guess PV is out for Rocks.
>

What complains about invalid kernel?

I think redhat does ship pv-on-hvm drivers with rhel/centos, so you
could enable those and get paravirtualized disk and network for the HVM
guest.

>    To set e1000 (virtualized Intel Gigabit NIC), I have to use the model
>    parameter in vif in HVM config? Will I get gigabit speeds instead of the
>    Fast Ethernet 100Mb/s speed with virtual Realtek NICs?
>

In a virtual guest those reported speeds don't mean much.. so even when the
realtek nic reports 100 Mbit/sec speed, it can go faster.

But yeah, model=e1000 and it will be reported as 1Gbit/sec.

Qemu e1000 emulation is faster than realtek emulation.. so that should be better.

-- Pasi

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel