Oh ho, thank you Pavel. So a HA cluster would be the base. From my research it seems that oVirt+Gluster is up-and-coming. Uses KVM as a base, which should give flexibility. I'd also like to find out what Amazon AWS uses for their ECC. Whether it's based on open-source or is entirely in-house. On Wed, Oct 16, 2013, at 10:11, Pavel Herrmann wrote: > Hey > > Im in no way an expert in the field, but i might be able to help you a > bit > > for a HPC cluster you would usually have a HA cluster for shared > resources > (like storage). then you would have a couple of compute nodes, > preferrably > with a fast interconnect (infiniband) > > the HPC part would be part of the application, be it MPI or hadoop or > whatever > > > On Wednesday 16 of October 2013 10:01:58 merc1984@xxxxxx wrote: > > And as I said several times in IRC, I am doing this for purposes of > > -learning-. How many times do I have to say it? > > > > I am doing this to learn about HP clusters. I am trying to get a handle > > on the landscape and the practical aspects of building a supercomputer. > > > > And I said this again in my first email here, that I am doing this to > > learn. > > > > Seems like you two don't have enough experience with this to advise, so > > please stand down. I'll figure out my own answers if this is the > > landscape. > > > > On Wed, Oct 16, 2013, at 9:30, Digimer wrote: > > > On 16/10/13 12:24, merc1984@xxxxxx wrote: > > > > I know. But keep in mind that RedHat's cluster is --High > > > > Availability--. > > > > > > > > What I am asking about is --High Performance-- as I say below. > > > > > > As was mentioned in IRC, you need to come up with a particular goal in > > > order for people to offer useful advice. What do you want to make go > > > faster? Answer that and you will start getting help. > > > > > > Cheers > -- http://www.fastmail.fm - Send your email first class -- Linux-cluster mailing list Linux-cluster@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/linux-cluster
