[Gluster-users] NUFA replicate-distribute

Anand Babu Periasamy ab at gluster.com
Sat Oct 31 09:52:18 UTC 2009

Whole purpose of NUFA is to give affinity to local disks. In a replicated mode, writes 
have to be written across servers. We implemented NUFA solely for HPC scratch processing.
For most work loads, plain distribute is recommended.

See this example below: You need local-volume-name option. Make sure you also include 
write-behind, read-ahead and io-cache translators for performance.

volume nufa
  type cluster/nufa
  option local-volume-name `hostname` # note the backquote, so 'hostname' output will be 
used as the option.
  subvolumes node01 node02 node03 node04

Fabio Bozzo wrote:
> Hi. I followed this reference: 
>> http://gluster.com/community/documentation/index.php/Mixing_DHT_and_AFR
> All works nice but I'd like to benefit of NUFA system too (new file created in the 'nearest' position to the author).
> How can I do?
> volume distribute
>   type cluster/nufa
>   subvolumes replicate1 replicate2
> end-volume
> Is this enough?
> Thanks.

Happy Hacking!
Anand Babu Periasamy
GPG Key ID: 0x62E15A31
Blog [http://unlocksmith.org]
GlusterFS [http://www.gluster.org]
GNU/Linux [http://www.gnu.org]

More information about the Gluster-users mailing list