[Beowulf] Building new cluster - estimate

[Mark Hahn]

>> vendors have at least list prices available on their websites.
>
> I saw only one vendor siliconmechanics.com that has online integrator.
> Others require direct contact of a saleperson.

the price of the cluster should be dominated by the price of a node,
and many sites offer web-configuration of nodes (HP, for instance).

>> "thermal management"?  servers need cold air in front and unobstructed
>> exhaust.  that means open or mesh front/back (and blanking panels).
>
> Yes, are there other options? Built-in airconditioning unit that will
> exhaust the hot air through a pipe to dump the air outside  computer room
> other than heating it?

there are racks with built-in heat-exchangers, yes.  I was mainly cautioning
against racks that are closed and have exhaust fans on the top, since they 
contradict the front-to-back airflow of servers.

>> wouldn't a 5100-based board allow you to avoid the premium of fbdimms?
>
> May be I am wrong but I saw only FB-DIMMs options and assumed that we need
> to wait for Nehalems for DDR3?

5100+ddr2 is perfectly viable.  fbdimms, after all, just a wrapper/extender
that introduces more latency with the claim of higher capacity (they contain
ddr2 or ddr3 inside the memory-buffer interface.)

but yes, if possible, I would wait for nehalem.  unfortunately, 
it's hard to guess when nehalem will actually ship for servers - 
recent rumors claim an early release, but I can't guess whether 
this is only for uni-socket machines or not.

>>  - WD Caviar 750 Gb SATA HD
>>>  :
>>> $110
>>
>> I usually figure a node should have zero or as many disks as feasible.
>
> We prefer to avoid intensive IO over network, therefore, use local scratch.

exactly, so more than one disk will provide better throughput.  remember
that one sata disk is about the same speed as 1 Gb interface.

>> HP is safe and solid and not cheap.  for a small cluster like this, I don't
>> think vendor integration is terribly important.
>>
> Yes, it is important. Optimized cost is what matters.

you asked for opinions: mine is that for small clusters, vendor integration
is not important because anyone can build a small cluster without much
effort.  vendor support is more of a necessity for large clusters.