580 likes | 749 Views
NOREX Peer-to-Peer Networking seminar Dave Powers – LRL R&D IT Andrew Kaczorek – LRL R&D IT Chris Chalfant – LRL R&D IT 07 October 2009. Cloud Computing inside R&D at Eli Lilly and Company. LRL IT Cloud Computing Agenda. Who are we? Where have we been (2008)? DEMO
E N D
NOREX Peer-to-Peer Networking seminar Dave Powers – LRL R&D IT Andrew Kaczorek – LRL R&D IT Chris Chalfant – LRL R&D IT 07 October 2009 Cloud Computing inside R&D at Eli Lilly and Company
LRL IT Cloud Computing Agenda • Who are we? • Where have we been (2008)? • DEMO • Where are we now (2009)? • DEMO • Where are we headed (2010)? • Q&A
Who is Eli Lilly and Company? • A heritage more than 130 years strong: company founded on May 10, 1876 • Headquarters located in Indianapolis, Indiana, U.S.A. • Approximately 39,140 employees worldwide • Approximately 7,486 employees engaged in research and development • Clinical research conducted in more than 50 countries • Research and development facilities located in 8 countries • Manufacturing plants located in 13 countries • Products marketed in 143 countries
Who we are… Part-time Server Architects Part-time HPC admins Part-time project managers Part-time cloud Admins Part-time Cloud evangelists Part-time Storage Architects Part-time Developers Our Background
Lilly IT: Key Business Challenges • Fixed costs • At the same time, moving to a FIPNET model: • collaboration • data sharing • software code/development sharing • computing resource consumption increasing • network access to data from anywhere at any time • security • Dependence on robust, reliable, scalable, flexible, secure infrastructure • Computing resource consumption increasing • Sense of urgency to change how we work
Our response to the challenge… • Tactically enable people to • Work: • Faster • Cheaper • More Efficiently • Pursue unconstrained models, simulations, hypotheses • Try bold ideas and approaches to solving business problems • Use technology as never imagined!
Where have we been (2008)? • “Cloud Computing” becomes official buzz-word • Defining “the cloud” • Layers (IaaS, PaaS, SaaS) • Vendors clamor to cloud • Amazon,Google, SalesForce • LRL IT pursue IaaS with AmazonAWS • Exposing possibilities • Real use cases • HPC, collaboration, data sharing
Where have we been (2008)? • “Cloud Computing” becomes official buzz-word • Defining “the cloud” • Layers (IaaS, PaaS, SaaS) • Vendors clamor to cloud • Amazon,Google, SalesForce • LRL IT pursue IaaS with AmazonAWS • Exposing possibilities • Real use cases • HPC, collaboration, data sharing
Defining Layers of “the cloud” Don’t care about writing code, just want the running application Software as a Service (SaaS) Don’t care about hardware, just want to login to environment and write/compile code Platform as a Service (PaaS) Rent the servers and storage on an “as needed basis” Infrastructure as a Service (IaaS)
Where have we been (2008)? • “Cloud Computing” becomes official buzz-word • Defining “the cloud” • Layers (IaaS, PaaS, SaaS) • Vendors clamor to cloud • Amazon,Google, SalesForce • LRL IT pursue IaaS with AmazonAWS • Exposing possibilities • Real use cases • HPC, collaboration, data sharing
Vendors clamor to cloud http://www.opencrowd.com/assets/images/views/views_cloud-tax-lrg.png
Where have we been (2008)? • “Cloud Computing” becomes official buzz-word • Defining “the cloud” • Layers (IaaS, PaaS, SaaS) • Vendors clamor to cloud • Amazon,Google, SalesForce • LRL IT pursue IaaS with AmazonAWS • Exposing possibilities • Real use cases • HPC, collaboration, data sharing
Where have we been (2008)? • “Cloud Computing” becomes official buzz-word • Defining “the cloud” • LAYERS (IaaS, PaaS, SaaS) • Vendors clamor to cloud • Amazon,Google, SalesForce • LRL IT pursue IaaS with AmazonAWS • Exposing possibilities • Real use cases • HPC, collaboration, data sharing
Real use cases • Proteomics – biomarker identification • 1000 cores • Collaboration - shared algorithms ( R ) • Pharmacogenomics • CloudBox – securely exchange data • Computational Chemistry due diligence • SAP sandbox • SugarCRM • VMC – self service
How We Consume Cloud Computing Depends on Who You Ask enduser developer scientist sysadmin management architect finance collaborator
How We Consume Cloud Computing Depends on Who You Ask enduser developer scientist sysadmin management architect finance collaborator “I want to be able to fulfill a variety of very specific infrastructure requests in record time”
How We Consume Cloud Computing Depends on Who You Ask enduser developer scientist sysadmin management architect finance collaborator “I want to be able to spin up custom environments on my own and tear them down quickly”
How We Consume Cloud Computing Depends on Who You Ask enduser developer scientist sysadmin management architect finance collaborator “I care minimally about the underlying environment, but I need to run some computationally intensive algorithms”
How We Consume Cloud Computing Depends on Who You Ask enduser developer scientist sysadmin management architect finance collaborator “I don’t care how it’s done; I simply want to interact with an application that is fast, reliable, and seamless”
How We Consume Cloud Computing Depends on Who You Ask enduser developer scientist sysadmin management architect finance collaborator “I want to easily prototype complex interconnected environments without managing technical details”
How We Consume Cloud Computing Depends on Who You Ask enduser developer scientist sysadmin management architect finance collaborator “I need to share data and applications between third parties with minimal lead time and friction”
How We Consume Cloud Computing Depends on Who You Ask enduser developer scientist sysadmin management architect finance collaborator “I need precise trending of upcoming costs, and I must decrease capital expenditures when possible”
How We Consume Cloud Computing Depends on Who You Ask enduser developer scientist sysadmin management architect finance collaborator “We need to accomplish more work per headcount, floortile, and dollar while maintaining high standards of quality”
Immediate Benefit of Cloud Computing: Forcing Us to Automate • Even extremely virtualized environments within • the enterprise are often managed like physical assets The dynamic nature of cloud computing requires considering servers, data, and applications as modular resources IT automation further up the stack is an enormous culture shift and requires strong technical leadership
Another Benefit: Transparency of Cost Like many enterprises, we struggle to define TCO and tie costs back to specific projects, groups, and services When using public cloud resources, every dimension of usage is broken down and presented in a periodic bill. While strategically important, the movement away from monolithic capital expenditures to small departmental charges is another cultural shift for us.
Where have we been (2008)? With whom have we been talking? • Legal • Security • Privacy • Quality/Compliance teams • Communications team • Procurement • IT groups (US and OUS) • Business areas • Vendors • Other Pharma companies (and non-Pharma enterprises)
Where are we now (2009)? • Delivering on IaaS with an eye toward PaaS and SaaS • Private cloud / Public Cloud / Hybrid Cloud • Building “Amazon-in-a-box” • Migrating 10 HPC applications to cloud in Q4/2009 • Leverage the cloud for data sharing • Continue exposing possibilities • Self Service Utility Computing (Q4/2009) • LRL IT Beta release of version 1.0
Introducing a New offering… • Delivery of VMC service • Virtual Machine computing? NO!!
Vending Machine Computing! Lilly VMC
Vending machine offerings today… Soda Books Dvds Cigarettes Noodles Candy Gum Beer Chips Coffee Ice Cream Milk
Why not a vending machine for virtual computers? • On-Demand • Self-Service • Ease of use • Cost friendly • Response time friendly • Variety of selections
Vending Machine Computing! DEMO 2
New Perspective of Time and Money • The cloud has enabled us to measure response time in minutes • The cloud has enabled us to measure costs in : • pennies / hour for compute • pennies / Gigabyte for storage
Pricing models – Let’s put it into perspective • Amazon AWS => $0.10 / hour • TerremarkvCloud=> $0.036 / hour • Let’s put it into perspective: • gumball from gumball machine ($0.25) => 7 hours • Candy bar from vending machine ($.70) => 19 hours • Bottle of soda from vending machine ($1.50) => 42 hours • Creative marketing ideas to foster creativity and trying new ideas…. • JetBlue => fly as much as you want for $599…..offer ends after 30 days • Lilly => compute as much as you want for $50….offer ends after 30 days • Ideas to foster creativity and trying new ideas • Starting today, it’s a different conversation with the business
Where are we headed (2010)? • The cloud to become: • Default end point for dev/test environments • Default environment for new HPC applications • Self service machine deployments to become the norm • Including associated metering, billing, reporting • Cloud environments become the infrastructure foundation • Private cloud, public cloud, hybrid cloud • Move up higher in the stack • PaaS • SaaS • Stretch goal => A beta version of a Lilly “App Store”
Interacting with the Cloud (2008-2009) Lilly – United States “The Cloud” Amazon rackspace Google Nirvanix xCalibre GoGrid
Interacting with the cloud with Orachestration (2010) Lilly – United States Orchestration Layer “The Cloud” Amazon rackspace Google Nirvanix GoGrid
2009 Applications HPC Self Service Private cloud Public cloud Virtual Private cloud Foundational Services
2010 Applications HPC Self Service Orchestrator Private cloud Public cloud Virtual Private cloud Foundational Services
Hybrid Clouds (2011) Google “Public Cloud” GoGrid Amazon Microsoft SkyTap SOA “Lilly Cloud” Europe China United States SOA India Singapore
Hybrid Cloud Orchestration (2012) Google “Public Cloud” GoGrid Amazon Microsoft SkyTap “Lilly Cloud” Europe United States China Orchestration Layer India Singapore
Implications for software developers • The bar is raised on security: • Encryption • In transit • At rest • Authentication • Identity • keys • Tightly coupled versus loosely coupled • Write once, run any where • Largely Distributed • High Scalability • High Availability
Changing our mindset What’s IN and What’s OUT? • Out… • Download, configure, make, make install • Enterprise licenses • The Infrastructure • CapEx (you bought it, you own it) • Many cooks in the kitchen • “Enterprise” Time • How much does it cost? • Status Quo • IN... • Fully functional VM appliance deployment • Utility-based licenses • The application • OpEx (pay-as-you-go) • Self Service/Vending Machine • Cloud Time • Transparency of Cost • Creativity and Innovation
How to get there? • Change our mindset • No longer only inside Eli Lilly’s data center • Move to Self Service • Drive to “cloud time” • Adapt the quality/compliance/security rigor to the technology ; not the other way around • Don’t lose the speed-up gains to traditional process approaches • Inspire people to see the possibilities • See the future not as a threat, but as an opportunity