3 years of OpenStack with Intel IT Das Kamhout – Principal Engineer, Cloud Architect @dkamhout Greg Bunce – Automation and Integration Lead Sridhar Mahankali – Cloud Architect
changing traditional thinking 100,000 Intel Employees 164 Intel Sites across 63 Countries 6,500 IT Employees inspire employees 59 IT sites globally service reliability 68 Data Centers 25% reduction with virtualization 150,000 Connected Systems IT is business 40,000 Handheld Devices Intel Con fl dential
Intel Data Center Pro fl le Intel has fl ve major groups currently driving individual data center requirements (DOMES): D Design § Supports chip design community Design Computing O Office § Supports typical IT and customer services General Purpose M Manufacturing § Supports fabrication and assembly E Enterprise § Enterprise applications supporting e Business and ERP S Services § External facing applications
Cloud: Experience from our Design Computing Grid IT Learnings 80% Utilization and Abstracted the hardware • an estimated $200M Value Abstracted the location • Service management • Service provisioning • $33M 78% 70% 2.31M 63% Of ffi oading Design Workloads 1.85M to Virtual Linux* Clusters 59% $25M 1.18M 0.99M Data Center Virtualization $0M 2006-1H 2006 2007 2008 1 Source: Intel IT internal analysis. Savings from DCV expected to deliver net present value over 8 years. Intel IT white paper: “Intel IT Data Center Solutions: Strategies to Improve Efficiency” http://communities.intel.com/docs/ DOC-4220 2 Source. Intel IT white paper on projected Intel net present value. “Realizing Data Center Savings with an Accelerated Server Refresh Strategy” http://communities.intel.com/docs/DOC-3489
Hosting Business Goals Velocity ty Velocity ty Idea to Production in <1 day <1hr for VMs Zero Down wnti time me Reduce Incidents ts “Always On” Scheduled Downtimes the norm for Apps/Services Grow w wi with th Flat t Budget Susta tain Operati tions Increase in Engineer:Server and TB Ratio Incr Increase Velo locit city, Zero Do Down wntim ime, Grow w wit with h Fla lat Bud udget
Server Landing Process (before Q4 2010) Requestor/ Customer Customer enters Reassess Request Backup setup Enter KCDB Server Request Customer Request in RADAR Requirements for VM if needed escalation info Fulfilled Accepts No No RADAR @http:// hosting.intel.com AM Updates Customer of Status Yes Notify Customer Address Decommission of Server Server Manager Request Account Availability Issues Review all Requests in Gather App and Server (AM) Enter forecast in Create HUM meeting. Assign server Requirements at Discovery Standard Implement Related Yes SHERPA/CPS Engagement requests (Virtual & Dedicated) Meeting with Customer; Request? EA Decommission Forecasting Tool Agreement (EA) to SERVER AM discuss EA/Costs No Yes 2012-2013 2009 2010-11 Sherpa/CPA Technologist Forecast No MAS Analyze Further and Design Yes Solution Solution Location (Customer, Validate Configuration Physical Possible? <30 minutes compute AM, Technologist Involvement) in SHERPA or Virtual? 90 days physical <3hrs virtual IPRO storage and network 24 days virtual 2 weeks for networks DIS IHS Assign LUNs; Validate Capacity Got Check Pick-up VM Create cutsheet for VM in SHERPA Capacity? Capacity Requests Capacity Mgmt Worksheet Procureme Cutsheet ADS Create IPRO Requests for Assign Existing or Pick-up Approved Dedicated nt IPRO Dedicated Server Landing Purchase Server Server Requests Operations Pickup/Create IPRO Assign to Site Request Network Install Configure Enroll VM in ISD Install Heartbeat Close Post Build Grant User Notify DC Next up Idea to Production Service in < 1 day and/or Cutsheet DC Ops Addresses OS Using Backup NIC Care Patching Monitoring for IPRO Verification Permissions AM Landing Request Representative (Primary/Backup) Altiris on all VMs (Sat 8-2) the VM Ticket
Private Cloud Economics Rent the Spike Predictability Performance Own the Base 7
Intel IT Cloud Strategic Direction Deliver the necessary changes in how we expose applications/data to improve end user productivity Drive the transformation to a large-scale automated Hybrid Cloud infrastructure Accelerate the transformation of the Enterprise IT industry to Cloud
History and Path to Open Cloud 9
Intel IT Cloud Quick History Ente terprise Private te Cloud Open Source Private te 2010 2010 Cloud 2012 2012 Design Grid since 1990’ 1990’s 13k VMs across 10 1.5k VMs across 2 60k servers across 60+ datacenters datacenters datacenters 75% of Enterprise Running cloud-aware Cloud’s Uncle Server Requests and some traditional apps 80% virtualized
Vision: Federated, Interoperable, Open Hybrid Cloud • Feder Federated: ted: IT manages one set of identities, IaaS, PaaS, and/or IaaS, P aaS, and/or Saas Saas authorizations, and set of security review Public Cloud Service Provider processes. Users get seamless integration with Public Cloud systems and apps. Service Provider • In Interoperable le: Standard service orchestration and management, enabling elastic operation and ffi exibility, while minimizing lock-in. • Op Open: n: Includes open source and open Enterprise Private Cloud standards. Common APIs and abstraction layers to rapidly consume cloud services among providers. App compone nents will will run run acr cross pub ublic lic and nd priv rivate clo cloud uds 11
Enterprise Adoption Roadmap - Path to Open Cloud Ecosystem Start Year 1 Year 2 Year 3 Year 4 Year 5 Consume mers Complex Simple SaaS Simple SaaS Hybrid SaaS End End Legacy Applications on dedicated Infrastructure SaaS User User Cloud Aware Cloud Aware Cloud Aware Apps Enterprise Apps Apps App Ap Legacy Apps Enterprise Dev Dev Legacy Apps Legacy Apps Legacy Apps Federated, Complex Interoperable, Compute and Open Cloud Open Cloud Ap App IaaS Private PaaS Hybrid PaaS Own wner Simple Simple Compute Compute IaaS IaaS IT Ops IT Ops Compute, Compute, Storage, and Storage, and Full Private Hybrid IaaS Network Network IaaS
Intel IT Pre-OpenStack – Private Cloud Gen 1 Start Year 1 Year 2 Consume mers End End Legacy Applications on dedicated Infrastructure User User Cloud Aware • Provides Self-Service to App Teams Apps Enterprise • Connect to ALL available infrastructure App Ap Legacy Apps Enterprise Dev Dev • $14M savings through resource pooling Legacy Apps • Internal Code for logic/gui Complex Compute Ap App IaaS Own wner Simple But… Simple Compute Compute • Cloud-aware app teams needed more IaaS IaaS • Too much technical debt to create full IT Ops IT Ops Compute, Compute, Storage, and IaaS Storage, and Network Network
Intel IT Post-OpenStack Private Cloud Gen2 Start Year 1 Year 2 Year 3 Consume mers End End Legacy Applications on dedicated Infrastructure User User • 2011 investigated all open Cloud Aware Cloud Aware Apps and proprietary solutions Enterprise Apps App Ap Legacy Apps Enterprise • Analysis led to decision: Dev Dev Legacy Apps Legacy Apps OpenStack for Private IaaS Complex • June 2012 online for Compute Ap App IaaS Private PaaS production cloud-aware apps Own wner Simple Simple Compute Compute IaaS IaaS But… IT Ops IT Ops Compute, Compute, Need a public cloud solution • Storage, and Storage, and Full Private Network Legacy apps need love too Network IaaS •
Intel IT OpenStack – Hybrid Cloud and the future Start Year 3 Year 4 Year 5 • Live Migration Enabled Consume mers • Single Control Plane End End Legacy Applications on dedicated Infrastructure • 2 POCs for Hybrid User User OpenStack in progress Cloud Aware Cloud Aware Apps Apps App Ap Very close to our year 5 goal Dev Dev Legacy Apps Legacy Apps Federated, Interoperable, and Open Cloud Open Cloud App Ap Private PaaS Private PaaS Own wner IT Ops IT Ops Full Private Hybrid IaaS IaaS
Our Choices and Architectural Implementation 16
Key Concepts • Abstract users from underlying Cloud providers while exposing key HW features • Support multiple cloud providers, both private and public • Common identity and entitlement services for reuse across interfaces • Open Source first, minimize proprietary API lock-in • Minimize internal technical debt, be part of the community to scale • Stay pragmatic, as we expand – not always 100% greenfield • Support cloud-aware and traditional apps
Technical Strategy (AS IS) On Premise Exclave Internal Network PaaS & DBaaS IaaS PaaS & DBaaS App Owner/ IaaS Firewa Developer wall Started in 2010 • Use our own capacity before paying an external • provider Public Clouds Intel IT at Service Provider size • Use public cloud for speci fl c purpose (SaaS, some IaaS) •
Technical Strategy (TO BE) On Premise Internal Network Exclave Firewa Orchestr PaaS & DBaaS trati IaaS App Owner/ tion Developer Burst Smart orchestration layer wall Move apps/data among clouds via policies • Public Clouds Deliver security, capacity and cost optimization •
Recommend
More recommend
