(Chris Forest and Adri Van Duin, co-chairs)
Accomplishments for Fall 2015/Spring 2016:
- Discussed a request from a faculty group for University support of a faculty-owned computer cluster. This resulted in a formal document with recommendations that was passed on through the RCCI Executive Committee to the VPR recommending either a reduction of the ICS-ACI pricing or, if not feasible, VPR support for faculty owned computer clusters. We are optimistic that the recommendations of the HPC WG and EC on this matter will directly lead to a substantial improvement in ICS-ACI pricing for the upcoming academic year.
- Invited ICS-ACI to provide a post-mortem on the LionX outage around Thanksgiving 2015. This resulted in an overview that was presented in the HPG WG and sent to the LionX user base. The HPC WG commended the devotion and technical expertise of ICS staff in the ICS handling of this outage. The HPC WG also offered recommendations for improving communications for any future outages.
- Inventoried HPC strategies in other universities and compared their situation with Penn State.
- Discussed the need for multiple hardware/cost/service models within ICS. We understand that ICS-ACI is now working to increase the number of service models available to researchers, incorporating inputs from the HPC WG.
- Assisted in the formulation of an HPC survey that will provide data critical to defining what hardware should be purchased as part of the upcoming the ICS-ACI Phase 3 computing environment. This survey was sent to faculty and researchers in April 2016 and received over 300 responses.
- Discussed the fate of decommissioned LionX-machines. This is more complicated that one would think and is an ongoing discussion.
- Formulated guidelines to ICS for the future decommissioning of the remaining LionX-machines and eventually ICS-procured hardware. Recommended operating machines after warranties expire with a reduced level of support (e.g., not replacing defective nodes) and providing users a minimum 6-month warning to the user-based of the to-be-decommissioned machine.
- Discussed Amazon AWS services and relevance for different types of academic research computing.
- Discussed NIST requirements for data handling – ongoing discussion.
- Discussed new Tower Road Data Center and its implications for HPC.
- Discussed connection between the HPC WG and ICS organizational structures.
Planned Work for Fall 2016/Spring 2017:
- Assist in interpreting the computing survey and formulating the plans for ICS-ACI Phase III procurement.
- Provide recommendations for current and future service options offered by ICS-ACI, e.g., What should be part of new service level agreements (SLAs)?
- Participate in communications between HPC WG and Data Center WG to project future power needs for research computing needs and thus timing of future phases of Data Center buildout.
- Discuss shutdown schedule of Lion-X and ACI clusters and develop communication plan for users
- Develop relations between our HPC working group and the ICS working groups.