[Gluster-devel] client config to access (mirrored) storage cluster

Daniel Maher dma+gluster at witbe.net
Wed Apr 2 11:31:24 UTC 2008


On Wed, 2 Apr 2008 16:23:44 +0530 "Anand Avati" <avati at zresearch.com>
wrote:

> > Ultimately, the big question i'm attempting to resolve is as
> > follows :
> > What is the best practise method to define client connectivity to
> > an HA Gluster cluster ?
> 
> dns round robin at the moment, HA translator when it is available.
> Have you investigated whether loading AFr on the client side works
> well for you? AFR handles failover as well.

Thank you for your response.

Out of curiosity, are there any documents related to the upcoming HA
translator available online ?

As for AFR on the client side - i haven't tried it, but in the time
between sending my original email and receiving your response, i read
over this (alternative) HA Gluster cluster document :
http://www.gluster.org/docs/index.php/Simple_High_Availability_Storage_with_GlusterFS_1.3

This document appears to suggest a scenario in which the fileservers
have no knowledge of each other, and it's up to the clients to write to
each of the fileservers (via afr or afr/unify).  In this fashion, the
clients are ultimately responsible for the mirroring functionality - is
this a correct interpretation ?


-- 
Daniel Maher <dma AT witbe.net>





More information about the Gluster-devel mailing list