kernel parameters for improving gluster writes on millions of small writes (long)

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 07/26/2012 11:47 PM, Harry Mangalam wrote:
...
> 
> So why doesn't the gluster native client do client-side caching like
> NFS?  It looks like it's explicitly refusing to be cached by the usual
> (and usually excellent) Linux mechanisms.
> What's the reason for declining this OS advantage on the client side
> while providing such a technically sweet solution on the server side?
> I'm at a loss to explain this behavior to our technical group.
> 

My understanding is this is a limitation of fuse moreso than glusterfs.
fuse currently fires off each write() it receives to the client fs
(gluster). FWIW, there is a fuse enhancement under development that you
can check out over on fuse-devel:

http://article.gmane.org/gmane.linux.file-systems/65661

I can't say whether that would solve your performance problems, but you
could certainly give it a try. I believe the intent is to sink writes
into the page cache (similar to NFS or a local filesystem) and send out
larger requests down to the fuse filesystem when writeback kicks in.

Brian


[Index of Archives]     [Gluster Development]     [Linux Filesytems Development]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Bugtraq]     [Linux OMAP]     [Linux MIPS]     [eCos]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux