[Info-vax] HPE and supercomputing.

[Scott Dorsey]

Stephen Hoffman  <seaohveh at hoffmanlabs.invalid> wrote:
>HPE was aiming at bigger server installations with their Apollo and 
>Moonshot series servers, and their "Synergy" "Compos[t]able" 
>"Hyperconverged" "Bladed" "Infrastructure" "BuzzPhraseReady" systems.
>
>That Apollo and Moonshot server sales focus seems to have largely 
>faded, and now it's seemingly all about the Cray and SGI high-end boxes.

It's going to be interesting.  We are a big customer for the SGI systems
and the Apollo servers, and HPE has done a few things to lock down the
Apollo servers that has made them problematic for us.  The SGI systems are
sold as cluster-in-a-box systems but you can throw out all the awful SGI
software and put generic Red Hat on them and they run nicely.

We won't touch the current Cray hardware (although we were once a very large
customer of theirs) because the Cray interconnect is proprietary and while
it may have lower latency than Infinband or Ethernet, it's not possible to
run the Cray clusters without the Cray software distribution and that means
paying through the nose for support.

The pricing models that Cray, SGI, and the traditional HPE systems use are
three very different models and it's going to be interesting to see HPE
try and unify them.
--scott
-- 
"C'est un Nagra. C'est suisse, et tres, tres precis."