Hi Deepak,
Ultimately, you're the only one who can know whether it works for your
exact workload or not. We recommend benchmarking your own workload.
That said, in general, if you avoid large numbers of small files, and
your hardware and network are sized appropriately, you should get good
throughput performance.
In HPC, a common problem is oversubscribing either the network or the
storage system. If you workload is I/O-heavy, you should spend probably
50% of your cluster budget on I/O and interconnect.
Regards,
Alex
On 02/09/2017 01:58 PM, Deepak Naidu wrote:> Folks,
>
>
>
> Wanted to get some inputs on the type of GlusterFS Volume is best suited
> for HPC workload which is throughput intensive. Anyone using GlusterFS
> in their env for HPC workload.
>
>
>
> I want to keep a balance of data usage & redundancy. I want to try
> erasure coded(dispersed) volume, not sure if its throughput intensive or
> not.
>
>
>
> PS: I don?t have any GlusterFS env running, still in the though process.
>
>
>
> --
>
> Deepak
>
> ------------------------------------------------------------------------
> This email message is for the sole use of the intended recipient(s) and
> may contain confidential information. Any unauthorized review, use,
> disclosure or distribution is prohibited. If you are not the intended
> recipient, please contact the sender by reply email and destroy all
> copies of the original message.
> ------------------------------------------------------------------------
>
>
> _______________________________________________
> Gluster-users mailing list
> Gluster-users at gluster.org
> http://lists.gluster.org/mailman/listinfo/gluster-users
>
--
Alex Chekholko chekh at stanford.edu