[Gluster-users] available space / performance / fault tolerance

[marianna cattani]

dear all,
I have 4 servers to realize the "block storage" for my OpenStack .
Which topology is the best to balance the three requirements in subject ?

For example , I can configure the nodes as a linked list , which allows me
to have a good fault tolerance :
two non-adjacent nodes can fail without interrupting service, I spend half
of the available space, but what about performance ?

Or if I try with dispersed volume in RAID5-like is it faster than a linked
list (obviously I gain space, but a single node could fail) ?
Which kind of parameters do I need to considered for this problem ?
Can I apply the same ideas that apply to a hardware raid (
http://en.wikipedia.org/wiki/Standard_RAID_levels) ?

Are there any other topologies that I can imagine with only 4 nodes (i.e.
RAID10-like with dispersed volume) ?

BR

M.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-users/attachments/20150405/03511d41/attachment.html>