Update LDM-572 on the Chilean DAC accounting for a more concrete approach in terms of machines etc.

From Jeff
 

1. ??? user compute nodes with a total of 300+ cores (number of nodes TBD based on cores/CPU, CPUs/node), max configurable memory, and LSP installed and networked with at least 2 x 1 Gbps NICs
2. Sufficient storage for 1500 home directories (local storage on compute nodes) totaling ??? TB/node (or use networked storage for home directories as well?)
3. Sufficient storage for DR2 catalogs and image files, with catalogs hosted in qserv, networked via multiple 10 Gbps connections to achieve an aggregate i/o throughput of at least ???.
4. Sufficient storage for 1 PB of user-generated data sets networked via multiple 10 Gbps connections to achieve aggregate i/o throughput of at least ???.
5. Network capabilities for the DAC will be based on single Top of Rack switches (not dual as formerly estimated).  Other non-DAC racks will have dual TOR switches.

 
Note that the above presumes all of the remaining BDC infrastructure identified in LSE-309 and listed in the table I sent last time, e.g. IDM, archivers/forwarders, etc. Those items, while not part of the Chilean DAC, will be integrated with the DAC as necessary (e.g Puppet/GSS, IDM, etc.).