This Wiki-based monograph is presented as part of a 9-month practicum project, which is a requirement of my Master of Business Administration degree with a Certificate in Sustainable Energy Systems. The MBA program is at the Bainbridge Graduate Institute, an innovative locus of study on existing and emerging sustainable business practices.
Research computing centers at academic institutions in the US face new and ongoing challenges. This work takes a broad look at some of those challenges, with an emphasis on the role of energy in high performance computing. It is done in the context of the leading organization for academic research computing and supercomputing in the US, the Coalition for Academic Scientific Computation (CASC).
The project seeks to elucidate the academic research computing environment as observed in literature, as offered by CASC members and others, and resulting from the author's own experiences. Tools are described to help interact with stakeholders, including systems analysis, budgeting and amortization, and some of the general language and concepts from business. Inasmuch as energy costs are major components of high-end computing operations, and a constraint on forward evolution of next-generation systems, energy is a constant theme.
This monograph and the associated spreadsheet are the major outcomes of the project. The project was developed from November 2013 until May 2014, under advisement by CASC members and affiliates, along with instructors and peers at BGI. The Table of Contents provides an overview of topics addressed.
This project has conceptual, temporal and target scoping limitations.
- Conceptual scope: We are concerned with business aspects of academic research computing in the US, with particular emphasis on energy. This necessitates a broad approach to research computing environments, and an understanding of research computing organizations, their people, and pertinent aspects of their environment. Energy emphasis includes practical engineering limitations of power dissipation for computing, financial aspects of electrical costs and the impact of such costs on return on investment (ROI), and sustainability aspects of where and how electricity is generated, and total embedded energy and carbon in computing equipment.
- Temporal scope: This project is intended as a snapshot, and does not attempt to fulfill a historical or retrospective function other than what is needed for clarity in a description of the present. Similarly, it does not attempt to look forward beyond those sections where planning for the next few years is described. An additional temporal scope limitation is that the project was due for completion by June 8, 2014, and might not be updated after that date.
- Target scope: The project is concerned mainly with the member organizations of CASC, and a small number of organizations that are not members but share characteristics. International considerations are of interest, but investigations into non-US institutions was limited. With over 3000 institutions of higher education in the US, the 75 or so CASC members represent under 1%. However, the CASC membership seems to be the majority of colleges or universities with substantial research computing capabilities, and is therefore judged by the author to be representative.
High Level Timeline and Milestones
The effort was spread across three academic quarters, with major milestones scheduled for the end of each quarter.
- First quarter (ended December 20, 2013): Project planning including communication plan and exit/completion strategy; stakeholder analysis; key resources analysis; industry analysis.
- Second quarter (ended March 21, 2014): completed sections on Energy, Physics, and Equipment Lifecycles; identification of ROI from this project to member organizations. Initial delivery of financial tools.
- Third quarter (ended June 8, 2014): Gathering of feedback from CASC members; investigations of industry business models in member organizations. Delivery of financial tools for budgeting, ROI, amortization and other purposes. Project completion; implementation of project exit strategy and any needed hand-offs.
The project outline is the Table of Contents on the main page for the monograph, and was adjusted as needed based on input from CASC members and other stakeholders.
Metrics and Measurement of Outcomes
The project sponsors will be apprised regularly of progress. Suggestions for adjustments to content will be discussed. The Table of Contents, in conjunction with the high level timeline (above) guided the effort. CASC members were be invited to visit the monograph and provide input on a quarterly basis. In the second quarter, CASC members were be solicited directly for their experiences in the business of research computing.
A number of Key Performance Indicators (KPIs) and associated metrics were developed by the author. Note, these are KPIs for this project, not for CASC members or other purposes. They will not all be measurable during the time of the project, due to natural delays in implementation and adoption. The CASC membership will be solicited to gauge the suitability, desirability, and reasonableness of these KPIs. Consistently with the "triple bottom line (TBL)" approach to business, metrics do not need to focus on revenues or profits. Emphasis areas include people, planet, and profit.
|Area of Influence||Indicator||Metric||Target||Goal: December 21 2013||Goal: March 20 2014||Goal: June 8 2014||Notes|
|Organizational||Industry analysis written and accepted||Deemed thorough and correct by project sponsors; evidence/stories of innovation by CASC members||100% acceptance||Draft communicated to CASC membership||Final version incorporated into final report||Industry to include research computing centers, commercial interests, government interests, and non-US research computing consortia|
|Social||KPIs developed||Deemed thorough and useful by project sponsors||100% of KPIs deemed appropriate||Feedback from CASC membership on draft version||Feedback from CASC membership on final version||Itemization of KPIs of potential interest for the CASC audience|
|Social||Utility as perceived by CASC members||Received feedback (structured & unstructured)||80% positive, 10% constructive, 10% critical||Initial feedback from sponsors||Feedback from CASC membership on draft version||Feedback from CASC membership on final version||Without some critical feedback, the final product is likely to be too cautious or tepid.|
|Financial; planet||Energy savings techniques adopted||Stakeholders report behavior change or plans based on energy analysis in the report||1MW saved 2014, ramping up to 10MW saved by 2017||Early assessment of current state of member operations||Initial reports of intent to change behavior||Reports of intent, and of change||Due to multi-year lifecycle installed systems, there is a necessary delay in adoption|
|Social||Research center articulation of social purpose||# of centers with articulated social purpose||Intent of 100% of CASC by June 8 2014||Initial description to stakeholders||Identification of early adopters and exemplars||Poll of CASC community||Industry analysis will include identification of articulated social purpose by CASC members|
|Social||Employee development plan at research computing centers||% of centers with articulated employee development plan||Intent of 100% of CASC by June 8 2014||Initial description to stakeholders||Identification of early adopters and exemplars||Poll of CASC community||Industry analysis will include identification of articulated social purpose by CASC members|
It is hoped that some elements of this monograph will be useful to some people. It has been continuously available, first as a working draft, since mid-October 2013. It was hoped that sponsors would be responsive to requests for feedback, and that the CASC membership and other stakeholders would provide timely input when requested. Because the author knows many CASC constituents personally, and is part of the research computing community, he was able to go directly to potential respondents to request a response. Feedback and input was, indeed, timely and useful.
That said, this project was designed to be resilient to a shortfall of input and to schedule slips. The author adjusted the timeline and expectations as circumstances warranted, while retaining focus on the major milestones and the expressed desires of the sponsors and other stakeholders. The Table of Contents, KPIs and monograph sections were adjusted as needed to reflect needed changes. Only major disruptions, such as an extended author illness, major new developments in electrical usage for digital computing, or shifts in mission for major stakeholders such as the National Science Foundation, would have required wholesale reconsideration of the project's structure and approach.