Benchmarking Unstable

Benchmarking Unstable

[Jacob Gorm Hansen]

hi,

we have some experiments running on a recent (less than a week old) Xen 
unstable. We are seeing dom0 HTTP performance (measuring using curl 
repeatedly fetching the same 1mb file from a nearby host over a gigabit 
link over HTTP1.1) that is three times worse than native Linux on the 
same machine.

The machine is a Pentium 4 and the Ethernet card is an Intel e1000. Xen 
has debugging turned off, and we are using the default scheduler. Only 
dom0 is running.

Is running benchmarks on Unstable considered a sane thing to do these 
days? Does anyone have ideas for how we can increase the performance in 
this setup?

Thanks,
Jacob


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

RE: Benchmarking Unstable

[Ian Pratt]

 

> we have some experiments running on a recent (less than a 
> week old) Xen 
> unstable. We are seeing dom0 HTTP performance (measuring using curl 
> repeatedly fetching the same 1mb file from a nearby host over 
> a gigabit 
> link over HTTP1.1) that is three times worse than native Linux on the 
> same machine.
> The machine is a Pentium 4 and the Ethernet card is an Intel 
> e1000. Xen 
> has debugging turned off, and we are using the default 
> scheduler. Only 
> dom0 is running.

If you're seeing a x3 reduction in performance from dom0 then something
is seriously broken. 

You might want to try 2.0.5 to see whether you get the same problem.
Looking at the number of interrupts per second might reveal something. 
 
> Is running benchmarks on Unstable considered a sane thing to do these 
> days? 

We've been doing benchmarks on unstable without problems.

Linux 2.6.11 (native) has some performance problems because of the
4-level page tables, but nothing that could cause a x3 slowdown.

Ian


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id\x14396&op=click

Re: Benchmarking Unstable

[Jacob Gorm Hansen]

Ian Pratt wrote:

> If you're seeing a x3 reduction in performance from dom0 then something
> is seriously broken. 

You are right, turned out debugging wasn't really off... The numbers we 
are seeing now are much better. There didn't seem to be any difference 
between stable and unstable performance-wise btw.

Thanks,
Jacob


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click