IT STRATEGY BOARD May 13, 2016
AGENDA > Call to Order > Research Computing — Network 20/20 Vision Update — Cloud — Consulting — Identity and Tier > Major Projects Review — HR/Payroll Modernization Update — UW Medicine EPIC Migration Project > IT Project Portfolio Executive Review > Wrap up 2
Research Computing Kelli Trosvig Vice President, UW-IT and Chief Information Officer 3
Network 20/20 Vision Update David Morton Director, NDT Mobile Communications, UW-IT 4
Since We Last Met Evaluated Options ˃ Technology BIG leaps and small steps — Gigabit Passive Optical Network (GPON) — Fiber to the desk — Wi-Fi everywhere — Wi-Fionly — Cellular/Wi-Ficonvergence — Routingcenter consolidation — Micro-power savings – devices sleep between packets ˃ Infrastructure and Standards — Cabling Cat6a — Number of cables pulled — Optimize comm closet size — Emergency power options — Outdoor hybrid fiber/power cable 5
Partners and Next Steps ˃ Consulted During Evaluation Phase ˃ Next Steps — Housing & Food Services — Complete Draft Findings presentation — Present to campus partners and solicit feedback — Capital Planning and Development > CPD, UWMed, HFS, CSE, ARTSI, UWB, UWT — UW Bothell > Other recommended partnerships/groups? — Adjust based on feedback — Computer Science & Engineering — Complete draft report/updated presentation — UW Medical Center — Report findings here in fall — WSU, Penn State, Princeton, Ohio State, WWU, Texas A&M, others 6
Questions or Suggestions? 7
Cloud Rob Fatland Director of Cloud and Data Solutions, UW-IT 8
Research premise ~ cloud value > Research should not be compute-limited > Data management, collaboration should be simple > Research computing is… — Perfunctory work, exploratory work, iteration > Wall-clock time is the primary resource — Lemma 1: Time cost/benefit and adoption barriers —Lemma 2: What about what we don’t know about? 9
The Molecules of Cloud Computing Three Examples from Today Eichler Lab Genetic Architecture of Autism $20k, 4 days (not 32 days) __________ 500 to Noble Lab 5000 etc. Genome-wide Association Studies Proteins Baker Lab Peptide-based Therapeutics (200) Cells (100) A x 25 = 1 Peptide protein A Left- handed complements… 10
Cloud Adoption 10 years > Three pennies gets you … (1 Hr / 1 GB-month) > Code “0—2%” (just the good bits) > Tim: Acceleration ‘89 to 1’ > Gaurav: Right-handed scaffolding > Tychele: Future impossible 5 years today Cloud is an element of research computing. Early adopters provide start- up effort; and they cut their costs and get the time back on their wall clock. And they get to ask and answer bigger questions. Cloud is part of the direction of research computing. Therefore we want to harmonize what we do ‘cloud’ with the rest of research computing. > Jupyter notebooks > API access to data > databases not spreadsheets > Frameworks not scratch software Hence: Consulting is the central idea for propelling 11 UW research computing forward.
Yes but what about… > Cost > Security > Reliability > Speed > Services 12
Research Computing Directions 1. Institutionalize scientific consulting at UW Next topic Next topic – ROI: ‘reverse consult’ story harvest > metrics 2. UW IT return to supporting research, starting with ‘cloud - capable’ – Double the staff of UW IT cloud computing for research 3. Democratize cloud access across the student population 4. Cloud adoption incentive programs: – Hardware retirement for r eciprocal cloud credit in lieu of ‘buy more servers’ – Proposal support, moderation of cloud research credit programs from vendors 5. Socialize cloud awareness – Deans conduct a ‘survey & exhort’ campaign – From each school or department: Cloud standard bearers – Existence-of and Practical training – Support cycle of participatory events: Cloud COP, hackathons, courses, incubators, … 6. IOT Grand Challenge 7. Implement Student-to-Research labor exchange with cloud certification 8. Growth model: UW campuses, medical; and partner universities 13
QUESTIONS 14
Consulting Bill Howe Associate Director, eScience Institute Rob Fatland Director of Cloud and Data Solutions, UW-IT 15
2005 - 2008 “All across our campus, the process of discovery will increasingly rely on researchers’ ability to extract knowledge from vast amounts of data… In order to remain at the forefront, UW must be a leader in advancing these techniques and technologies, and in making [them] accessible to researchers in the broadest imaginable range of fields.” In other words: > Data-intensive science will be ubiquitous > It’s about intellectual infrastructure and software infrastructure, not only computational infrastructure 16 http://escience.washington.edu
Long Tail of Research Data [src: Carol Goble] GenBank PDB ChemSpider UniProt CATH, SCOP Pfam (Protein Structure Spreadsheets, Notebooks Classification) Local, Lost 17
Where Do You Store Your Data? Lewis et al 2011 My computer 87% External device (hard drive, thumb drive) 66% Department-managed server 41% Server managed by research group 27% External (non-UW) data center 12% Department-managed data center 6% Other 5% 0% 20% 40% 60% 80% 100% src: Conversations with Research Leaders (2008) src: Faculty Technology Survey (2011) 18
How Much Data Do You Work With? Wright 2013 19
eScience Research Consulting circa 2010… ~2 FTE z Technical staff Jeff Gardner Bill Howe David Beck Erik Lundberg Chance Reschke 20
eScience Research Consulting 2016 ~8.5 FTE Research Faculty Research IT Data Scientists (fully supported) Rob Fatland Jake VanderPlas Dave Beck Bernease Herman Valentina Staneva Ariel Rokem Director of Cloud and Director of Research, Director of Research, Data Scientist Data Scientist Data Scientist Data Solutions Physical Sciences Life Sciences BS, Stats Ph.D., Applied Ph.D., Senior Data Science Ph.D., Astronomy Ph.D. Medicinal was SE at Amazon Mathematics Neuroscience Fellow Chemistry, and Statistics PhD Geophysics Biomolecular Structure & Design Research Scientists (partial support) Dir. Ethnography Joe Hellerstein Bryna Hazelton Vaughn Iverson Andrew Gartland Brittany Fiore-Gartland Senior Data Science Fellow Research Scientist Research Scientist Research Scientist Ethnographer IBM Research, Ph.D., Physics Ph.D., Oceanography Ph.D., Biostatistics Ph.D Communication Microsoft Research, Google (ret.) 21
WRF Data Science Studio 22
Building a Research Consulting Portfolio > 2010: Embedded > 2011: Ad hoc lab visits and 1-hour meetings > 2012: Tutorials, events, user groups > 2014: Incubator > 2015: Office hours > 2015: Data Science for Social Good > 2016: Rob Fatland 23
go to them Duration of Engagement come to us per FTE: Joint 0-2 Research annually 2010-present 0-2 2010-present Embedded annually 1-2 2014-present Incubator; DSSG annually 25-30 2011-present Door-to-Door; Lab Visits annually 50+ Office Hours 2015-present annually # of engagements 24
go to them Duration of Engagement come to us Joint Research 2010-present 2010-present Embedded 2014-present Incubator; DSSG 2011-present Door-to-Door; Lab Visits Office Hours 2015-present # of engagements 25
26
go to them Duration of Engagement come to us Joint Research 2010-present 2010-present Embedded 2014-present Incubator; DSSG 2011-present Door-to-Door; Lab Visits Office Hours 2015-present # of engagements 27
Data Science Kickoff Session: 137 posters from 30+ departments and units 5/9/2016 Bill Howe, UW 28 28
go to them Duration of Engagement come to us Joint Research 2010-present 2010-present Embedded 2014-present Incubator; DSSG 2011-present Door-to-Door; Lab Visits Office Hours 2015-present # of engagements 29
Incubation Program Overview > Quarter-long, in-Studio projects, engagement two days per week — 4-6 concurrent teams: Network effects among cohort beyond 1:1 interactions — Each team is ~50% project lead + ~50% eScience FTE — Structured, time-bounded engagement ensures progress (and an exit strategy) — Feels like a course: “I have incubator today, so I can’t go do XXX” > Two-page proposals describing a shovel-ready problem, the science, and how a solution can generalize to other groups > Participation from faculty, grad students, masters students, and staff http://escience.washington.edu/get-involved/incubator-programs/ 30
31
32
Assessing Community Well-Being Third-Place Technologies Optimization of King County Metro Paratransit Computer Science & Engineering Predictors of Permanent Housing for Homeless Families Bill and Melinda Gates Foundation Open Sidewalk Graph for Accessible Trip Planning Electrical Engineering 33
go to them Duration of Engagement come to us Joint Research 2010-present 2010-present Embedded 2014-present Incubator; DSSG 2011-present Door-to-Door; Lab Visits Office Hours 2015-present # of engagements 34
Office Hours 35
Recommend
More recommend
