[Gluster-users] PLEASE READ ! We need your opinion. GSOC-2014 and the Gluster community

Paul Robert Marino prmarino1 at gmail.com
Mon Mar 17 15:37:00 UTC 2014


well how about underlying volume monitoring,
right now if an underlying volume goes bad Gluster doesnt know about
it but if it could detect issues in the file system and or physical
disk it would be useful in the future for automatically migrating the
data off the effected brick(s)
this would allow Gluster to be used on flat disks without underlying raid.

By the way I wouldn't get too excited about what Isolon is doing with
SSD cache nodes. Its effectiveness varies depending on the size of the
files you are dealing with. It really only makes the handling of small
files faster but while handling large files larger files it actually
can slow the cluster down. This is because spinning disks doing
sequential reads of large files are still faster than SSD's. Isolon
has some great sales people with a lot of technical knowledge about
their products, but for obvious reasons they tend to avoid telling you
their known weak point till you are already buying from them and even
then they are hesitant to tell you unless you have a serious problem
which could be solved by buying more hardware from them.
SSD's handle random access of small files well. Isolons traditionally
have poor performance with small files but very good performance with
large files. what the SSD cache node is really useful for is if you
have a lot of small files on your Isolon cluster (which you shouldn't)
it significantly increases the speed on those small files and reduces
the overhead caused by accessing small files on the storage nodes.
That said if you have mostly large files and few to no small files
which is what Isolon storage clusters were designed to do the SSD
cache node is just a pointless expense.

On Thu, Mar 13, 2014 at 6:38 PM, Ted Miller <tmiller at hcjb.org> wrote:
>
> [snip]
>
> 2) We have a recurring issue with split-brain solution. There is an entry on
> trello asking/suggesting a mechanism that arbitrates this resolution
> automatically. I pretty much think this could come together with another
> solution that is file replication consistency check.
>
> Anything to improve split-brain resolution would get my vote.
>
> Ted Miller
> Elkhart, IN
>
>
> _______________________________________________
> Gluster-users mailing list
> Gluster-users at gluster.org
> http://supercolony.gluster.org/mailman/listinfo/gluster-users



More information about the Gluster-users mailing list