...
We had planned to intersperse storage purchases in with compute nodes. The new cluster architecture is relying on Lustre as a shared file system, and also to provide scratch space for batch. Such an upgrade was anticipated and the 170 TB existing space can be doubled by adding trays to the existing servers for about $60k.
Purchase Options
Anchor Options Options
| Options | |
| Options |
This proposal is to expand the bullet cluster with combined funds from
PPA general, ATLAS, and Theory. This would double our existing parallel
file system size (173->346TB) and add XX cores.
This proposal is to expand the bullet cluster with combined funds from PPA, ATLAS, and Theory. This would double our existing parallel file system size (173->346TB) and add either 1649 or 1904 cores depending on which option we choose. The first option is to provision infiniband (IB) in all nodes and add IB switches to allow additional future expansion of the IB network. Because of the IB network topology allowing future expansion implies a jump in the number of core switches from 4 to 8. The second option would split the cluster into IB and non-IB parts with the ATLAS nodes being non-IB. Note the pricing below is based on several different quotes that would have to be refreshed. Hence the pricing is approximate and hopefully not low-balled. The details are:
...
Option 1: Expand to 18 fully populated chassis with all-IB and future expansion capability
...
- 6 full chassis @91k => 546k
- 7 blades w/IB added to existing empty slots @4.817k => 33.72k
- 4 IB switches with cables @10.5k => 42k
- 2 60x2TB disk trays with controllers @34k => 68k
Total is $690k for 1648 cores and storage expansion.
Gross bullet cluster core count would then be 4608 (all IB)
costs are:
- ATLAS: 60 blades @4.9375k => 296.25k (960c)
- Theory: 91k + 7*4.817k => 124.72k (368c)
- PPA: 268k (320c)
Note the PPA cost/core is bad because it includes the storage expansion and IB infrastructure.
...
Option 2: Expand to 15 full IB chassis and 4 non-IB chassis
...
- 4 full non-IB chassis @79k => 316k
- 3 full IB chassis @91k => 273k
- 7 blades w/IB @4.817k => 33.72k
- 2 60x2TB disk trays with controllers @34k => 68k
Total is $691k for 1904 cores and storage.
Gross bullet IB cores is then 3840.
Gross bullet non-IB cores is 1024.
costs are:
- ATLAS: 4 @79 => 316k (1024c)
- Theory: 91k = 7x4.817k => 124.72k (368c)
- PPA: 269k (512c)
...
Add-on to either option:
...
We could get new GPU servers (kipac's are old!) which are equivalent to bullet blads with ~5000 gpu-cores for ~10k each. So we could top off to 300k if we got 3 of these. We (PPA) do need to replace or existing GPU "system" that is hosted by kipac.