Hi Sam,

Thanks for the extended report. I am in the middle of the iVDGL annual 
report, so I will look at this in detail later. Ruth is also free to comment.

Thanks,
Paul


At 07:45 PM 6/19/02, Lee Samuel Finn wrote:
>Dear Paul and Ruth:
>
>As Paul, Albert and I discussed last month, until iVDGL personnel are
>on-board at Penn State I will be reporting once a month via e-mail on
>progress and status of iVDGL activities at Penn State. This is the first
>such report.
>
>The overall focus of activities at Penn State have been on recruiting
>personnel and scoping the LIGO Data Analysis requirements on a Penn State
>greenfield tier-2 center.
>
>On the personnel front, we have two positions we intend to have filled by
>Fall 2002: these are a systems analyst level position and a postdoc level
>position. The systems analyst will be the principal systems administrator
>for the cluster and also support integration and operations of LDAS and
>grid/globus tools at Penn State. We seek to fill this position from the
>computer science community; however, the scarcity of qualified personnel at
>the budgeted salary may require that we turn back to the physics community
>for this position. The postdoc level position, which has been filled, will
>provide the interface between the systems development work and the LIGO data
>analysis community: i.e., support the porting of analysis codes to the PSU
>tier 2 site, and the use of grid/globus tools by the physics data analysis
>community. The postdoc offer was accepted June 1 with a start date of
>October 1.
>
>The Penn State situation is different then other iVDGL greenfield sites, in
>that there is an existing software system - LDAS - that it must support in
>order to fulfill its mission of servicing LIGO data analysis. Consequently,
>in consultation with the LIGO Lab, we have identified as a high priority
>maintaining the same basic configuration as the LIGO Laboratory tier-2
>facilities (LIGO Hanford Observatory, LIGO Livingston Observatory, and LIGO
>MIT). This configuration is discussed in the attached aide memoir, which
>summarizes the outcome of discussions with Lab personnel responsible for the
>design of the Lab tier-2 systems and the LDAS software system.
>
>Since the LIGO Lab tier-2 centers are part of the iVDGL tier-2 framework
>this requirement should not be in conflict with any iVDGL facilities
>requirements, which necessarily respect the existing centers at the LIGO
>Labs and UWM. Indeed, this choice should minimize the FTE drain that would
>be involved in implementing and supporting LDAS on yet another system
>architecture.
>
>In putting PSU forward as a tier 2 site I teamed with Vijay Agarwala,
>Director of Computing and Information Systems at Penn State, who has
>extensive experience in the development, deployment and operations of
>specialized computing clusters for numerically intensive computing work.
>Following upon the discussions with the LIGO Lab on the basic architecture
>of the LIGO Lab systems Vijay and I have settled on the purchase of a small
>number of nodes that will allow us to develop experience in supporting LDAS
>for production and make a few basic hardware choices for the full-sized
>cluster. With this experience, and consultation with the LIGO Scientific
>Collaboration Computing Committee and iVDGL Facilities Committee, we will
>proceed to the development of the full cluster, which will mirror in
>architecture a LIGO tier-2 center.
>
>The initial node purchases will include
>
>Eight 2.4GHz intel/1GB/36GB disk: compute part of LDAS cluster
>Two 2x2.4GHz intel/2GB/36GB disk: front-end to compute part of cluster,
>data-conditioning unit
>
>Two 2x2.4GHz intel/4GB/5x73GB disk: for testing intel solution for data
>server and database server
>
>Two Sun 280R (dual proc packed fiber channel disk): for testing sun solution
>for data server, database server
>
>While testing the intel solution for data and database server, we can use a
>280R for the front-end to the entire system (runs globus and other external
>connectivity services: everything else is behind it on a private net); while
>testing the sun solution, we can use one of the intel boxes as the
>front-end.
>
>Sun prices are currently being negotiated; intel hardware described above
>has already been negotiated at $46K.
>
>Best,
>
>Sam
>
>
>--
>Lee Samuel Finn                                  off: 814-863-9605
>Director, Center for Gravitational Wave Physics  fax: 814-863-9608
>Center for Gravitational Physics and Geometry
>Professor, Departments of Physics, Astronomy and Astrophysics
>The Pennsylvania State University
>