[Beowulf] NFS HPC survey results.
laytonjb at gmail.com
Sun Jul 24 10:43:29 PDT 2016
Thanks Bill and thanks Doug! While the sample size is small there are some
interesting things in the data. Just my 2 cents (probably worth alot less
- It looks like alot of people are just using Linux Servers and not
appliances (perhaps moving toward more of the open-storage model).
- It's also interesting that people aren't seeing NFS timeouts like they
used to. I remember a number of years NFS timeouts were very common and
people really trashed NFS as a result.
- The NFS options are also interesting. One thing I thought was cool was
that a few people are using "async". Using async mounts can improve
performance if you have a copy of the data somewhere. Basically it becomes
a scratch file system.
- Looks like /home is the place to use NFS :) But people are using it
other places (e.g. /usr/local).
- It's really cool that almost 50% of the respondents are using NFS for 65+
nodes (24% above 512).
- About 65% of the respondents are using Ethernet for the network.
Very interesting results.
On Fri, Jul 22, 2016 at 8:05 PM, Douglas Eadline <deadline at eadline.org>
> With Bill's permission, there is now an HTML version
> on Cluster Monkey
> > Many thanks for all the responses.
> > Here's the promised raw data:
> > https://wiki.cse.ucdavis.edu/_media/wiki:linux-hpc-nfs-survey.csv
> > I'll summarize the 26 results below. I'll email similar to those that
> > asked.
> Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing
> To change your subscription (digest mode or unsubscribe) visit
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Beowulf