Hi,
I doubt that dispersed volume will be faster as it needs to 'encode' some data during the writes.
Actually, if you don't need any redundancy and you work with large files -> you can test sharding xlator. Usually it's used for VMs and in your case the files will be sharded into small pieces (shards) spread among multiple bricks. Give it a try on a test volume first.
About the brick layout , I prefer to use 1 HW raid = 1 brick. For fast disks like NVMEs -> 1 NVME = 1 brick.
WARNING: ONCE SHARDING IS ENABLED, NEVER EVER DISABLE IT !
Best Regards,
Strahil Nikolov
On Sun, Nov 28, 2021 at 3:59, Patrick Nixon<pnixon@xxxxxxxxx> wrote:________Hello Glusters!
I've been running a multi-node single brick per node distributed array with the bricks being between 6 and 14TB each and getting okay performance.
I was reading some documentation and saw distributed dispersed as an option and was considering setting up a test array to see if that improved the performance. I don't need replicas / redundancy at all for this array, just bulk storage.
My question, primarily, is about how to layout the bricks across six nodes with the ability to add additional nodes/drives as necessary.
Option 1:Single Brick Per Node
Option 2:
Multiple Bricks Per Node
- Bricks a consistent size (1T each, left over disk as it's own brick)- Bricks a fraction of the total disk (1/4 or 1/2)
Thank you for any suggestions/tips (links to additional documentation that would help educate me are welcome as well).
Community Meeting Calendar:
Schedule -
Every 2nd and 4th Tuesday at 14:30 IST / 09:00 UTC
Bridge: https://meet.google.com/cpu-eiue-hvk
Gluster-users mailing list
Gluster-users@xxxxxxxxxxx
https://lists.gluster.org/mailman/listinfo/gluster-users
________ Community Meeting Calendar: Schedule - Every 2nd and 4th Tuesday at 14:30 IST / 09:00 UTC Bridge: https://meet.google.com/cpu-eiue-hvk Gluster-users mailing list Gluster-users@xxxxxxxxxxx https://lists.gluster.org/mailman/listinfo/gluster-users
