About Glusterfs and data replication

[Avi Greenbury]

jiten jha wrote:

> What is the use of GlusterFS and its meaning why we use.

It's like a more scalable form of NFS. Essentially, the point is to
provide highly-available network storage - you can configure multiple
servers to serve the same volumes such that if one server goes down all
operations from their clients are redirected to other servers. You
might also configure it such that there is a server at each
geographical location, each kept in sync with each other. Clients then
only connect to their geographically local server, with the appearance
of one global server.

It's used a lot where there are many hosts needing to share access to
one resource, specifcally where high-avaliablity is a requirement. It's
the de-facto standard means, for example, of serving the same content
across multiple availability zones in Amazon's cloud.

> I have 4 PC. what i did in a 2 pc i have install HAproxy and
> Heartbeat and other 2 pc apach ( HTTP) . All 4 pc connect to the
> switch. My question is I am using 2 pc as a server both pc have 500
> GB HDD so how to replicate data in both server HDD. What is the
> process to replicate data in both server . We do not have extra
> storage device .

You might find it easiest to use something like csync there. Or, if
it's just two machines sharing the device, OCFS atop DRBD would be
good. GlusterFS is probably a good deal more complex than you need, and
if you've consistently low network latency (which I'd expect with a
switch such as yours) DRBD should be fine.

-- 
Avi