14. The
60’s
The
70’s
The
80’s
The
90’s
The
00’s
From centralized to decentralized, collaborative to independent
and right back again!
The
10’s
Mainframes VAX
The
PC
Beowulf Clusters Central
Clouds
100% 60% 0% 40% ??? %
SHARING
~
0Mbit
~ 1Mbit ~ 10Mbit ~
1000
Mbit
~ 10,000 Mbit
Bigger, better but further and further away from the scientist’s lab
15. Ask a
Question Hypothesize Predict
Experiment /
Test Analyze Final Results
The Scientific Method
Test and Analyze stages
require the most time,
compute, and data
16. Ask a
Question Hypothesize Predict
Experiment /
Test Analyze Final Results
The Scientific Method
Any improvements to this
cycle yield multiplicative
benefits
17. A Challenge Across Industries
— 3 of Top 5 Insurance
— 6 of Top 8 Pharmaceutical
— 2 of Top 3 Banks
— 2 of Top 3 Genomics Sequencing
— 1 of Top 2 FPGA
18. Utility HPC in the News
WSJ, NYTimes, Wired, Bio-IT World BusinessWeek
26. We make software tools to easily orchestrate complex
workloads and data access across Utility HPC
Today is a survey of use cases…
10,600 instance
Life Science
Molecular
Modeling
600 core
Manufacturing
Nuclear Power
Plant for safety
simulation
Genomic
Analysis
RNA for
Stem Cells
37. Before:
Trade-off compute time vs.
accuracy
Now:
Accurate analysis, fewer false
negatives, faster
Initial
Coarse
Screen
Higher
Quality
Analysis
Best
Quality
Process for Drug Design
Higher
Quality
Analysis
Best
Quality
38. Big 10 Pharma
Built 10,600 instance cluster
($44M) in 2 hours, ran
40 years of science
in 11 hours for $4,372
42. Earlier Drug Design
Novartis discussed at BioIT2012
— Needed
— Push-button Utility Supercomputer for molecular
modeling
— Created
— 30,000 core run across US/EU Cloud (AWS)
— 10 years of compute in 8 hours for $10,000
— Found 3 compounds now in the wetlab as a result
43. — Capacity is no longer an issue
— Hardware = software
— Testing (error handling, unit testing, etc.)
e.g. Cycle spent ~$1M dollars on AWS over 5 years
— The only way to do this is to automate
Lessons learned
48. We don’t’ know what they’re
running, but it has “Safety”
49. 600-core CAD/CAM
3 Quarters of a year wait became 3 weeks
Site
Data
Corporate
Firewall
3 Weeks instead
Of 3 Quarters
Secure
HPC
Cluster
TBs FS
External Cloud
~600 CPU cluster
Scheduled
Data
Engineer
50. Survey of Use Cases
þ Drug Design
þ CAD/CAM
þ Genomics
…
51. Gene Expression Analysis
Morgridge Institute for Research
Run holistic comparison of all 78 terabyte stem cell
RNA samples to build a unique gene expression
database
Make it easier to replicate disease in petri dishes w/
induced stem cells
53. 1 Million compute hours,
115 years of computing in
1 week for $19,555
54. Gene Expression Analysis
Morgridge Institute for Research
— Cluster details
— 5,000 to 10,000 cores for a week
— Very long individual analysis were check-pointed =
Spot instance usage possible
55. Survey of Use Cases
þ Drug Design
þ CAD/CAM
þ Genomics
…
57. Ask a
Question Hypothesize Predict
Experiment /
Test Analyze Final Results
The Scientific Method on Utility HPC
Yield “Better”, “Faster”
Research for less $