March Meeting: GlusterFS (Aaron Toponce)

Aaron Toponce aaron.toponce at gmail.com
Wed Mar 6 13:10:12 MST 2013


On Wed, Mar 06, 2013 at 12:14:33PM -0700, Steve Meyers wrote:
> GlusterFS is setup using servers and clients, where clients use the FUSE 
> driver to communicate with the servers. Servers communicate with each 
> other, and are setup via exporting "bricks" to the cluster. It uses 
> eventual consistency, versus strong consistency, for data integrity.

So, I've heard from a number of presenters, from a few conferences, where
clustered filesystems are the topic, that GlusterFS is an eventually
consistent filesystem. Well, after submitting this to Steve, I started
thinking more about it, and I've come to the conclusion that I should
clarify:

    * Replication within the quorum is strongly consistent.
    * Geo-replication is evenually consistent.

Basically, "eventual consistency" means that all updates are eventually,
seen, but may not be available immediately, whereas "strong consistency"
assures that all updates are seen immediately after they have been
committed. Asynchronous writes are eventually consistent. Synchronous
writes are strongly consistent.

At any event, see you at the meeting.

-- 
. o .   o . o   . . o   o . .   . o .
. . o   . o o   o . o   . o o   . . o
o o o   . o .   . o o   o o .   o o o
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 519 bytes
Desc: not available
Url : http://plug.org/pipermail/plug/attachments/20130306/471cd67b/attachment.bin 


More information about the PLUG mailing list