[Gluster-users] Use case sanity check

[Ashley Large]

Hi

 

I have been hunting around for up to date info on my use case but most is
outdated so im not sure if still valid. I have about 500 web sites that I
want to host in AWS. I need to survive AWS outages so am considering Gluster
in cross region replicated setup. 

 

I had looked before at Gluster but keep coming across warnings about it
being ill suited to web serving due to slow reads on small files. The worst
thing I could do (besides lose all customer data!) is move from owned and
happy physical hardware to Gluster backed cloud, find slow loads of sites
and then have to unpick the solution again.

 

Also I would like to use same solution for Postfix which again is lots of
small files. With Postfix though there are also lots of small file writes
which seems to be another problem with Gluster.

 

I want the scalability and redundancy of Gluster but these things are
worrying. The sites are mainly low traffic but there are a few with major
peaks and some with consistent high volume. Mail is probably around 1500
pop3/imap users.

 

Also, at beginning I mentioned using cross region replicated setup. I
suspect this may have a network overhead that could have a big io impact?
Assuming Gluster is suited to Apache/MX usage, perhaps Geo replication is a
better option for me with the slave as hot standby?

 

Any help would be great thanks.

Ash

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20150205/b7aa97e5/attachment.html>