On Wed, 13 Jun 2012, Andrew Cathrow wrote:
What does your deployment look like - it's be great to understand
how you're using it - eg. Linux guests only, Windows, size of deployment - do your
users use poweruser portal etc.
I have 2 deployments I am working on with two different companies.
The first is two clusters (SJC and NYC) each with 20 boxes PXEBooted to
a backend NetApp. All guests are Linux (Centos, RHEL, and Fedora). One of
the unique things with the clusters is they use read only linux and NFS
root.
The 2nd is still in design stage, but is starting out with 3 clusters
(East / West US and Europe) each with 20 boxes. Each server is based on
white label Xeon E5-2687W, 128GB RAM, 10GBase-T, and 12 SATA disks on a
hardware RAID 6 controller. This is also going to use shared NFS read only
root with PXEBoot from NetApps.
I have been running GlusterFS on and off for 5 years and are looking at
running GlusterFS on all the raw 300TB RAID6 in each server. Each server
will be a brick, two servers with distribute for redundancy, 10
distribute pairs then will be unified. I have been trying to use this for
raw image store for years, but its just not ready, so we will use NetApp
for that, however GlusterFS will be very handy for tier 2 bulk storage.
GlusterFS and Quantum are certainly in the works, but in a separate
thread I'd love to hear feedback on the API what the issues/enhancements are.
Would you be able to post some initial comments on that?
APIs are great today on the backend, we are looking at the ability to
modify the front end so we can tie it into our portal allowing users to
manage their VMs seamless with the other services we offer.
<>
Nathan Stratton
nathan at
robotics.net
http://www.robotics.net