[Gluster-users] Best practices?

Joe Landman landman at scalableinformatics.com
Tue Jan 24 18:13:36 UTC 2012

On 01/24/2012 01:07 PM, Whit Blauvelt wrote:

> So is the preference now that even for workloads _not_ involving huge files,
> XFS is better? For non-huge-file systems is Ext4 more likely to break, or
> suffer in performance speed?

ext4 has some significant serialization in its journaling and other 
pathways.  Its great for boot drives.  Not so great when you have many 
simultaneous processes hammering on it.

Someone posted results that literally spoke for themselves last year as 
a response to a statement I made to that effect.  With 8 simultaneous 
readers/writers, ext4 was taking something like 2x (going on memory 
here, so don't take this as precise) the time that xfs was to perform 
the operations.  With 16 readers and writers and up it gets very pronounced.

xfs is designed for parallel IO workloads.  ext4 isn't, and it shows 
under load.

Aside from this, carving a storage system into 16TB chunks is a terrible 
thing to do to a large file system capable unit.  ext4 (if you follow 
the previous link) still really doesn't do 16TB+ stably.  Things break. 
  It will eventually get there, but its not there now.  xfs has been 
doing large file systems for more than a decade.

Joseph Landman, Ph.D
Founder and CEO
Scalable Informatics Inc.
email: landman at scalableinformatics.com
web  : http://scalableinformatics.com
phone: +1 734 786 8423 x121
fax  : +1 866 888 3112
cell : +1 734 612 4615

More information about the Gluster-users mailing list