I’d agree with Nux and use VM for management server. This works well and lends nicely to scaling up and out as needed.
I use compute/storage nodes with glusterfs and shared mount point but this could be overkill or waste of storage capacity if you don’t want or need the added redundancy. On Mar 26, 2014, at 11:57 AM, Talk Jesus <c...@talkjesus.com> wrote: > Yes I meant RAID10, not 1 :) Sorry about that. > > As for local storage, so you're "local" as in it's own dedicated server, > outside the HV (CPU/Memory) server correct? > > -----Original Message----- > From: Nux! [mailto:n...@li.nux.ro] > Sent: Wednesday, March 26, 2014 11:55 AM > To: users@cloudstack.apache.org > Subject: RE: New to CS, some prep questions > > On 26.03.2014 15:34, Talk Jesus wrote: >> Thanks for the response. I actually forgot to include one serve for >> actual storage. However, would it be more simplified and easier to >> just have a the storage server act also as the CPU/Memory (power) >> server too? Does CloudStack have some config ability to claim a server >> as an HA server (high availability)? > > You can use local storage, I would actually recommend it if you want SSDs, > but there is no way to achieve HA with local storage, you need it shared. > > >> Server #2 >> Purpose: Hypervisors (CPU + Memory + Storage) >> Qty: 1 plus 1 for failover (HA) >> Dual Hexacore CPUs >> 128GB RAM >> 12 x 1TB SAS (or SSD) RAID1 >> *above drives is for main storage, I know is pricey, but it's only for >> an example* > > I'd go for RAID10 in this case. > > Be advised, all this is IMHO, YMMV. > > -- > Sent from the Delta quadrant using Borg technology! > > Nux! > www.nux.ro >
