[CentOS] Optimizing NFS over loopback

Wed Aug 19 03:55:44 UTC 2009
Ruslan Sivak <russ at vshift.com>

Eugene Vilensky wrote:
>> Things are working, but seem pretty slow.  I'm getting about 10mb/sec.
>> I was wondering where the bottleneck could be?  Is there a way to
>> optimize the NFS service for better throughput?
>>
>> I am accessing over 127.0.0.1
>>
>>     
>
> Are the server and client coming from the same back-end spindles (I
> realize this is localhost)?  How many spindles are involved?
>
> If you are reading/writing to the same disk(s) expect seeking to eat
> up the vast majority of potential throughput unless you have a total
> pool of sufficient IOPS.
> _______________________________________________
> CentOS mailing list
> CentOS at centos.org
> http://lists.centos.org/mailman/listinfo/centos
>   
One server is a single drive and a second server has two 300gb 
velociraptors mirrored on a Dell Perc 5/i. 

Either way I should be getting higher speeds.  I'm getting about over 
100MB/s when testing /dev/sda3 with hdparm. 

Is there a test I can do to eliminate the disks as a bottleneck?  Maybe 
set something up in RAM?

Russ
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.centos.org/pipermail/centos/attachments/20090818/66b6ebb7/attachment-0005.html>