Faster, More Scalable Computing in the Cloud
Pavan Pant, Director Product Management
Using the Cloud for Infrastructure On Demand
CloudSwitch Proprietary & Confidential2
0.0%
27.3%
18.2%
9.1%
36.4%
9.1%
0%
1-5%
6-10%
11-25%
26-50%
>50%
Response %
Source: Insight Pharma Reports
% of Life Sciences R&D Informatics Budget Devoted
to Cloud Computing in 3YearsNeed massive
computing power
Used to scaling resources
Bear huge costs & delays of provisioning
internally
Shared resource environment offers economies of scale
The Market is Predicted to Have Significant Usage within 3 Years
Apps Ripe for Cloud Computing
Which Apps?• Next-generation DNA
sequencing– Pattern recognition– Data mining
• Molecular modeling & simulation
• Protein docking
Why?• Burst/peak scale out• Improving the
application lifecycle• Collaboration
CloudSwitch Proprietary & Confidential3Source: “Cloud Computing in Life Sciences,” Insight Pharma Reports, April, 2010; and Pharma Consultant
“Often we want to take the data that we have, marry it up with data that’s publicly available – could be big genomic data sets – look at all of that collectively, and then extract value from that. So they’re very bursty. They really peak. It’s lots of data and then you’re done.”
– Michael Heim, CIO, Eli Lilly
Early Adopters
• Eli Lilly:– 64-machine cluster using
Amazon EC2– Completed sequence processing
in 20 minutes vs 12 weeks– Cost: $6.40– Plan to have up to 10 HPC
applications “cloud enabled” by end of the year
CloudSwitch Proprietary & Confidential4
http://www.sramanamitra.com/2010/10/08/boundaries-between-hpc-and-cloud-computing-vanishing / http://www.expresspharmaonline.com/20101031/market01.shtml
“For us, it's pipeline, pipeline, pipeline. Anything we can do to further our knowledge, get products into the pipeline, and develop those more quickly, is crucial to us. It's hard to underestimate the value of letting scientists work at their own pace.” – Michael Heim, CIO, Eli Lilly
• Pfizer’s Biotherapeutics and Bioinnovation Center:– Used EC2 to develop & refine models in antibody-antigen docking runs– Shortened the process from days to hours
Changing Pharmaceutical Research Landscape
CloudSwitch Proprietary & Confidential5
Reduce time to discovery &
development
Reduce operational costs and capex
Increasingly complex data sets
& processing requirements
Growing collaboration and data sharing
What’s Needed to Make the Cloud Work
CloudSwitch Proprietary & Confidential6
Orchestration layer
More high-compute resources
More streamlined procurement
Enterprise-level implementations
Flexibility
Security
Ease of deployment/transparency with data center
Cloud resource provisioning
CloudSwitch Proprietary & Confidential7
CloudSwitch Product Architecture
Cloud 2
Customer Data Center
App 1
VIRTUAL MANAGEMENT/CONTROLS
VIRTUALIZED STORAGE
Cloud 1
CloudSwitchInstance
(CSI)
CLOUD ISOLATION TECHNOLOGYTM
CloudSwitchInstance
(CSI)
CLOUD ISOLATION TECHNOLOGYTM
Components work cooperatively to simplify, secure, and manage operating environments in the cloud
ENC
RYPT
ED T
UN
NEL
App 5
App 2
App 3App 3
App 2
App 5
VIRTUALIZED STORAGE
VIRTUALIZED STORAGE
App 2
App 3
App 2
App 3
DATA CENTER SERVICES:DNS, LDAP, Identity, Infrastructure…
App 4
CloudSwitchAppliance
(CSA)
FIR
EWA
LL
Use Case: Bio Informatics in the Cloud
Data Center (Internal)
Firewall Administrators
Cloud Compute Cluster
Scientists & Researchers
Data Center LAN
CloudSwitchInstance
(CSI)
CloudSwitchAppliance
(CSA)
SecureConnection
Data Center LAN
Data Server
QueueMaster
Compute 1 Compute 2 Compute n,000
ProvisionServer
Compute LAN
Data Source 1 Data Source 2
Compute Job Submission
8 CloudSwitch Proprietary & Confidential
CloudSwitch HPC Scenario
• Large Pharma with 1000 Cores in Amazon EC2– Created 500 compute node clones (1000 cores) in ~30 minutes– Provisioned all nodes via network boot (PXE) in the cloud in 45-60 minutes– Using Sun Grid Engine v6.2 as the queue master– Using Rocks v5.4 as the front-end “control” server – Started with 48-hour test for a high performance bioinformatics workload – Goal is to establish a more permanent footprint in the cloud
• Elasticity and Significantly Lower Capital Expenditure– Compute nodes are brought up when needed and shut down after the
compute process was finished– Total cost of less than $10,000 to run 1,000 cores in Amazon for 48 hours– All done securely and seamlessly using CloudSwitch as the management
control plane for the compute nodes and the Rocks and SGE environment
CloudSwitch Proprietary & Confidential9
CloudSwitch Confidential
Use Cases & Benefits of the Cloud for Healthcare
• Two Common Use Cases1. Cluster capacity for burst/peak demand
• Scale-out for research and informatics usage with data center control2. Dev/test environments in the cloud
• Offload from production gear• Enable self-service and scale testing• Bring back on-prem for production
• Benefits– Elasticity– Reduce Ongoing Costs – Process Complex Data Sets Via Horizontal Scaling in the Cloud
11/5/201010