Utility HPC: Right Systems, Right Scale, Right Science
-
Upload
chef-software-inc -
Category
Technology
-
view
1.342 -
download
0
Transcript of Utility HPC: Right Systems, Right Scale, Right Science
![Page 1: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/1.jpg)
Utility HPC: Right Systems, Right Scale,
Right Science
Jason Stowe, CEO @jasonastowe, @cyclecomputing
![Page 2: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/2.jpg)
I’m here to recruit you, for a cause
![Page 3: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/3.jpg)
We believe utility access to compute power
makes impossible science, possible.
![Page 4: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/4.jpg)
Dynamic, utility access to compute power
is as important as uptime
![Page 5: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/5.jpg)
(that’s why coded infrastructure is critical)
![Page 6: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/6.jpg)
Skeptical? Flickr: Tourist on Earth
![Page 7: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/7.jpg)
In prior years (today?)
Researchers/engineers waited for computing
![Page 8: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/8.jpg)
For the horsepower
![Page 9: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/9.jpg)
For the place to put it
![Page 10: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/10.jpg)
For it to be Configured..
Flickr: vaxomatic
![Page 11: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/11.jpg)
Yesterday, high performance engineering, science clusters
were…
Too small when you need it most,
Too large every other time.
![Page 12: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/12.jpg)
The Innovation Bottleneck: Researchers/Scientists/Engineers
Forced to size questions to the infrastructure you have
![Page 13: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/13.jpg)
Multi-‐tenant systems create float capacity That is critical to innovation
![Page 14: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/14.jpg)
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% ??? %
SHARIN
G ~ 0Mbit ~ 1Mbit ~ 10Mbit ~ 1000 Mbit ~ 10,000 Mbit
Bigger, better but further and further away from the scientist’s lab
![Page 15: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/15.jpg)
Ask a Question Hypothesize Predict Experiment /
Test Analyze Final Results
The Scientific Method
Test and Analyze stages require the most time,
compute, and data
![Page 16: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/16.jpg)
Ask a Question Hypothesize Predict Experiment /
Test Analyze Final Results
The Scientific Method
Any improvements to this cycle yield multiplicative
benefits
![Page 17: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/17.jpg)
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
![Page 18: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/18.jpg)
Utility HPC in the News�WSJ, NYTimes, Wired, Bio-IT World BusinessWeek
![Page 19: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/19.jpg)
To accelerate science, we need automation
![Page 20: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/20.jpg)
![Page 21: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/21.jpg)
Management Software
CC1/CCG Instances EBS S3
Shared FS
EBS
Utility HPC Cluster -‐ Scales to 50,000+ cores -‐ Data Scheduling -‐ Workload portability
Data & Application
Aware Movement
Traditional Scheduler
Massive Scale Based upon workload
Secure, HPC Cluster
User
HPC Reporting &
Audit
![Page 22: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/22.jpg)
50,000-core CycleCloud Using Chef and AWS
ChefConf 2012
![Page 23: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/23.jpg)
10,600-instance cluster against cancer target
ChefConf 2013
![Page 24: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/24.jpg)
Created in 2 hours Configured with Search,
with Data bags
![Page 25: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/25.jpg)
one Chef 11 server
![Page 26: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/26.jpg)
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
![Page 27: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/27.jpg)
Dynamic, utility access to compute power
is as important as uptime
![Page 28: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/28.jpg)
Why?
![Page 29: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/29.jpg)
#1: “Better” Science =
“Answer the question we want to ask”, not constrained to what fits
on local compute power
![Page 30: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/30.jpg)
#2 “Faster” Science =
Run this “better” science, that would have taken
months or years in hours or days
![Page 31: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/31.jpg)
Survey of Use Cases þ Drug Design þ CAD/CAM þ Genomics …
![Page 32: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/32.jpg)
Life Sciences & Compute? C
ompu
te
Data/Bandwidth
Genomics
Molecular Modeling
CAD/ CAM
All Sample Analysis
Proteomics Biomarker/
Image Analysis
Sensor Data Import
Creating fake Charts, with Fake Data
![Page 33: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/33.jpg)
Why is this important?
![Page 34: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/34.jpg)
(W.H.O./Globocan 2008)
![Page 35: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/35.jpg)
~2 million Type 2 diabetics, ~200k Type 1
![Page 36: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/36.jpg)
Every day is crucial and costly
![Page 37: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/37.jpg)
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
![Page 38: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/38.jpg)
Big 10 Pharma Built 10,600 instance cluster
($44M) in 2 hours, ran 40 years of science
in 11 hours for $4,372
![Page 39: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/39.jpg)
Most Recent Utility Supercomputer server count:
![Page 40: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/40.jpg)
AWS Console view:
![Page 41: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/41.jpg)
Cycle’s view of this cluster:
One Chef 11 Server
![Page 42: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/42.jpg)
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
![Page 43: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/43.jpg)
� 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
![Page 44: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/44.jpg)
Servers are not house plants
![Page 45: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/45.jpg)
Servers are wheat
![Page 46: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/46.jpg)
Survey of Use Cases þ Drug Design þ CAD/CAM þ Genomics …
![Page 47: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/47.jpg)
Nuclear Power Plant simulation
![Page 48: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/48.jpg)
We don’t’ know what they’re running, but it has “Safety”
![Page 49: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/49.jpg)
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
![Page 50: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/50.jpg)
Survey of Use Cases þ Drug Design þ CAD/CAM þ Genomics …
![Page 51: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/51.jpg)
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
![Page 52: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/52.jpg)
78 TB of Stem Cell RNA
![Page 53: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/53.jpg)
1 Million compute hours, 115 years of computing in
1 week for $19,555
![Page 54: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/54.jpg)
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
![Page 55: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/55.jpg)
Survey of Use Cases þ Drug Design þ CAD/CAM þ Genomics …
![Page 56: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/56.jpg)
Code can accelerate Science
![Page 57: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/57.jpg)
Ask a Question Hypothesize Predict Experiment /
Test Analyze Final Results
The Scientific Method on Utility HPC
Yield “Better”, “Faster” Research for less $
![Page 58: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/58.jpg)
Dynamic, utility access to compute power
is as important as uptime
![Page 59: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/59.jpg)
I’m here to recruit you, for a cause
![Page 60: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/60.jpg)
Contribute to Chef. Make the community better.
And you will help Cycle make impossible science,
possible.
![Page 61: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/61.jpg)
2013 BigScience Challenge
$10,000 of free computing to science benefitting humanity
2012 winner: 115yr Genomic analysis
Enter at: http://cyclecomputing.com/big-science-challenge/enter
![Page 62: Utility HPC: Right Systems, Right Scale, Right Science](https://reader034.fdocuments.us/reader034/viewer/2022052619/5558240ed8b42a5e468b5118/html5/thumbnails/62.jpg)
Thank You! Questions?