Investing in compute nodes, storage, or specialty infrastructure in an HPC is often advantageous for many researchers. These advantages can become more apparent when comparing the time and capital a researcher and their cohorts can expend when setting up, operating and later requiring support for self-managed research computing resources. Faculty and other University researchers may wish to purchase nodes to be placed in our HPC cluster using sponsored research funds, startup funds, or other sources of funding. Inquiries into our HPC Investment Program may be made here through our service portal.
Advantages to the Investor
Node purchases provide investors with preferred, on-demand access to their purchased computing resources.
Priority access is granted to investment related project(s), and this priority access is linked to their purchased resources.
Researchers may find they are in need of specialty hardware that is currently unavailable on the cluster. Investment purchases allow for users to have this specialty hardware integrated into the cluster to facilitate their specific area of research.
UW ARCC provides system administrators for investment equipment service and maintenance.
Investors nodes are granted a secure, climate controlled location in the UW IT Datacenter for their purchased equipment.
Users can access cluster storage provided and maintained by ARCC. This includes home, project, and gscratch directories.
Investor nodes are connected to high speed network infrastructure allowing users quality access to data and HPC computing resources.
As a researcher you would not need to investigate all components required to make your hardware and software solution work for you. ARCC does this for you.
The workload required added when adding hardware to an already configured cluster is far less when compared to the resources required to maintain smaller scale, discrete equipment. We may offer comparable services performed by knowledgable, on-site staff. It is likely worth your time to ask us if you aren’t sure whether your research related needs fall within our offerings. We can help!
Advantages to the University
Centralized node purchasing through UW Advanced Research Computing Center offers multiple benefits to the University and to your Department/Research Group:
Investing in our cluster and available resources increase UW’s overall HPC resource utilization for more efficient use of university-wide research computing dollars.
Overall node purchasing and sharing offers scaling benefits by reducing university-wide costs associated decentralized HPC infrastructure, maintenance, equipment, and support staff. As a researcher you would not need to investigate all components required to make a server work.
HPC Condominium Model
Our Beartooth cluster follows a condominium-style model of resource management. Investors, which are individual researchers or groups of researchers, work with us at ARCC to purchase compute nodes and storage. These resources are then installed in our cluster and administered by the ARCC system administrators.
When the investor-purchased resources aren’t in use by the investor defined users, they will be made available to the community. This type of arrangement is referred to as a backfill queue. However, this general access to investor-owned resources can be immediately revoked when the investors wish to use their resources. The job will automatically be re-queued to ease the burden on users. Investors are also given priority on the backfill queue.
Idle compute resources are made available to the general user community through a shared backfill queue that includes all compute nodes. However, this general access to resources can be preempted at any time by investors when they want to use the resources they have purchased.
If you wish to investigate purchasing nodes or know what you want and are ready to purchase nodes for use in the cluster, please submit a request at this link: https://arccwiki.atlassian.net/servicedesk/customer/portal/2/group/15/create/53. An ARCC representative will contact you to discuss details regarding the purchase. Once you decide to commit to a purchase we will work with you regarding Service Level and Capacity Management to develop a service level agreement (SLA) applicable for the equipment purchase (equipment, cost, service period, etc.).
While investment purchased equipment is subject to primary use by the investor for the full term of the service agreement, ARCC retains ownership of the equipment through it’s lifecycle. While Investment buys priority access for the investor, it does not buy node ownership.
The standard Service Level Agreement period for a compute or GPU node investment is 5 years. Other periods may be considered on a case-by-case basis including a proposal or award performance period.