U N C L A S S I F I E D LA-UR Taking the initiative on virtualization Anil Karmel Solutions Architect Network & Infrastructure Engineering Production Systems June 1, 2009

U N C L A S S I F I E D LA-UR Slide 1 Outline How Los Alamos National Laboratory implemented virtualization to reduce their carbon footprint and consolidate data centers across their campus Leveraging virtualization to supplement your DR/BC plan Identifying “low hanging fruit” for your lab’s green initiatives while achieving a substantial ROI Moving computing from the desktop to the data center to enhance your lab’s security.

U N C L A S S I F I E D LA-UR What is Virtualization? Slide 2 Low Utilization of Resources High Utilization of Resources

U N C L A S S I F I E D LA-UR How Los Alamos National Laboratory implemented virtualization to reduce their carbon footprint and consolidate data centers across their campus Slide 3

U N C L A S S I F I E D LA-UR Slide 4 What Challenges Did We Face? Application and Citrix Servers ~300 Intel-based HP ProLiant and Dell servers – Primarily 3 rd /4 th Generation HP ProLiant DL360 servers Enterprise Approach to Imaging, Patching, Backups – Allows for standardized management of heterogeneous servers Hosting Application and Citrix Environments – 32+ Web Applications – 10,000 hits/day – 50 Citrix Servers – 70 applications – Lotus Notes/Domino, WebSphere, SharePoint, Project, SQL Server, Exchange and others

U N C L A S S I F I E D LA-UR Slide 5 What Challenges Did We Face? Challenges Decommission Legacy Hardware Consolidation of Existing Infrastructure Need to Rapidly Provision New Environments Integration of IT Environments within LANL

U N C L A S S I F I E D LA-UR Slide 6 What Challenges Did We Face? Business Case GapAlignment Servers distributed in multiple data centers Reduction in number of servers will enable consolidation into fewer data center facilities Limited data center floor space, power, and cooling resources Reduction in number of servers will reduce data center requirements Need to retire unsupported platforms Unsupported platforms can be replaced by virtual servers Slow server provisioning cycle- time New environments provisioned without purchasing and installing additional physical servers Limited hardware for disaster recovery Environments can be restored onto different hardware. Hardware replaced by new virtual servers and servers retired during future refresh cycles can be utilized for DR site.

U N C L A S S I F I E D LA-UR Slide 7 VMware Capabilities VMware Virtual Infrastructure 3.5 VMotion – Move live virtual servers dynamically between ESX hosts Distributed Resource Scheduler – Manages ESX server workload based on system resources by automatically moving virtual servers High Availability – Automatic restart of virtual machines on other ESX servers in case of ESX host failure Resource Pools – Ability to aggregate, deploy and manage server resources based on preconfigured criteria (category, environment, customer, etc)

U N C L A S S I F I E D LA-UR Slide 8 VMware Capabilities VMware Virtual Infrastructure 3.5 VMware Update Manager – Push software updates to ESX hosts, VM’s and templates VMware vCenter Site Recovery Manager – Automates the failover and recovery process to meet RTO and RPO requirements – Create recovery plans for different failover scenarios – Perform non-disruptive tests of failover and recovery – Requires Array-Based Replication VMware vCenter Lifecycle Manager – Automates the entire lifecycle of a machine from creation to retirement

U N C L A S S I F I E D LA-UR Slide 9 VMware Architecture Consolidation onto Virtual Platform Hardware Configuration – Thirteen (13) HP ProLiant DL585 servers – Dual-core AMD Opteron processors (4) – 32GB memory – Multiple network and fiber channel cards – HP EVA 8000/8100 SAN environment (119TB total) – Shared Production VMware EVA 8000 – 27TB – Shared DEVL/BETA VMware EVA 8000 – 45TB – Shared DR EVA 8100 – 47TB – Active SAN-SAN Replication between EVA8k’s and EVA8100 – Near-line Storage for Disaster Recovery

U N C L A S S I F I E D LA-UR Slide 10 VMware Architecture Consolidation onto Virtual Platform Monitoring / Consolidation Planning – PlateSpin PowerRecon – Ability to gather Windows/Linux physical server utilization metrics – Metrics used to identify physical candidates for consolidation Physical to Virtual (P2V) Technology – PlateSpin PowerConvert – Ability to rapidly migrate physical machines into a virtual environment – Migrate physical machines running on disparate hardware to standard hardware – Create weekly images of physical servers into inactive VM’s for DR Backup Technology – Vizioncore ESX Ranger with VMware Consolidated Backup – Ability to backup entire VM for DR purposes

U N C L A S S I F I E D LA-UR Where Are We Today? Consolidated Infrastructure Statistics – 210 total VM’s running in a 13 Server VMware DRS/HA cluster – Average resource utilization of cluster is 50% – Anticipating a 40:1 consolidation ratio (dependant on load) – 105 physical servers decommissioned – 3 data centers retired Return On Investment (ROI) – Initial projection called for ROI to be achieved by end of FY ’08 – Actual ROI achieved by March 2007 (9 months) Plan to evaluate VMware vCenter Site Recovery Manager and vCenter Lifecycle Manager Slide 11

U N C L A S S I F I E D LA-UR Leveraging virtualization to supplement your Disaster Recovery / Business Continuity plan Slide 12

U N C L A S S I F I E D LA-UR Supplementing your DR/BC Plan The Current State of (Physical) DR DR services tiered according to business needs Physical DR is challenging – Maintain identical hardware at both locations – Apply upgrades and patches in parallel – Little automation – Error-prone and difficult to test TierRPORTOCost IImmediate $$$ II24+ hrs.48+ hrs.$$ III7+ days5+ days$

U N C L A S S I F I E D LA-UR Slide 14 Supplementing your DR/BC Plan Opportunities with a Virtual Solution Consolidation / Integration of Infrastructure Rapid Deployment of New Technologies Disaster Recovery Capability – Server images inherently portable and not reliant upon underlying hardware – Ability to move freed-up hardware to off-site location for subsequent DR hosting – P2V servers that need to remain physical for use in a DR scenario – Virtual hardware can be automatically configured – Test and failover can be automated (minimizes human error) – Costs are lowered, and the quality of service is raised Diverse Platform Support – Microsoft Windows Family – Linux and Solaris (x86 based)

U N C L A S S I F I E D LA-UR Slide 15 Supplementing your DR/BC Plan Opportunities with a Virtual Solution VMware Site Recovery Manager – Automate the failover to your DR site based on pre-defined recovery plans – Ability to execute non-disruptive tests of recovery plans – Control access to recovery plans with granular role-based access controls – Requires implementation of array-based replication technology and integration with SRM by storage vendor

U N C L A S S I F I E D LA-UR Slide 16 HP White Paper: HP disaster tolerant solutions using Continuous Access for HP StorageWorks Enterprise Virtual Array in a VMware Infrastructure 3 environment

U N C L A S S I F I E D LA-UR HP White Paper: HP disaster tolerant solutions using Continuous Access for HP StorageWorks Enterprise Virtual Array in a VMware Infrastructure 3 environment Slide 17

U N C L A S S I F I E D LA-UR Identifying “low hanging fruit” for your agency’s green initiatives while achieving a substantial ROI Slide 18

U N C L A S S I F I E D LA-UR Identifying low hanging fruit Utilization Rates On average: – Typical server use approximately 10% of it’s computing capacity – Typical desktop use approximately 20% of it’s computing capacity Slide 19

U N C L A S S I F I E D LA-UR Identifying low hanging fruit Recommendations CDW-G E2IT report 2008 – Virtualize as much as possible – Servers: Reduce the number of physical servers, racks, switches and cabling which lowers power consumption accordingly – Desktops: Hosting desktop computers and applications centrally in a data center and removing computers from employee desks can greatly reduce power consumption Employ monitoring tools to capture actual utilization – e.g. PlateSpin PowerRecon Decommission multiple physical endpoints to enhance green computing and achieve ROI Slide 20

U N C L A S S I F I E D LA-UR Where Are We Today? LANL Green Computing Energy Savings $1.435M Cost Avoidance / Savings to date – $700,000 Cost Avoidance – $735,000 Cost Savings Recipient of NNSA Best in Class Pollution Prevention Award for Server Virtualization Slide 21 Servers (kWh/yr) 200 Physical Servers 13 Physical Hosts Energy Savings Direct696,10756,822639,285 Indirect255,40020,847234,533 Total951,50777,669873,838

U N C L A S S I F I E D LA-UR Moving computing from the desktop to the data center to enhance your agency’s security Slide 22

U N C L A S S I F I E D LA-UR Moving from the desktop to data center Traditional PCs Benefits of Traditional PCs – Individual desktops for every user Limitations of Traditional PCs – Costly to maintain, upgrade and support – IT must have personnel in the field – No control over data residing remotely (e.g. laptops) – Difficult to maintain standard desktop images – Time-consuming to deploy or redeploy desktops Slide 23

U N C L A S S I F I E D LA-UR Moving from the desktop to data center Server-Based Computing Benefits of Server-Based Computing – Upgrade, patch, backup desktops in a single location – Secure confidential information in a secure data center – Deliver new desktops more quickly – Improve control over desktop images Limitations of Terminal Services & Blade PCs – Desktops aren’t always isolated – Application compatibility problems – Costly to dedicate a system to each user (e.g. Blade PCs) Slide 24

U N C L A S S I F I E D LA-UR Moving from the desktop to data center Desktop Virtualization Benefits of Desktop Virtualization – Easier to manage desktops with centralized encapsulated files – Deliver just enough server to desktop users – Move hardware independent desktops seamlessly among servers – Complete isolated desktops with no application changes – Desktop disaster recovery not previously available – Reduce / eliminate the need to refresh physical desktops Slide 25

U N C L A S S I F I E D LA-UR Moving from the desktop to data center Security Benefits Centralized management – Application and OS Upgrades – Patch Management Potential reduction of endpoints on your network – Reduce risk of a single unpatched system compromising your enterprise Slide 26

U N C L A S S I F I E D LA-UR Virtual Desktop Solutions Citrix XenDesktop Citrix XenDesktop – Uses PortICA to communicate from client to virtual desktop – Citrix Provisioning Server – Multiple base image share a common base disk – Ability to patch a single base disk and update multiple machines Slide 27

U N C L A S S I F I E D LA-UR Virtual Desktop Solutions VMware View VMware View – Uses RDP to communicate from client to virtual desktop – View Composer – Multiple base image can share a common base disk – Reduce storage costs by up to 90% – Ability to patch a single base disk and update multiple machines – Allows for persistent and non-persistent desktops – Offline Desktop Support (experimental) – Ability to check out an image to use on a remote endpoint (i.e. laptop) Slide 28

U N C L A S S I F I E D LA-UR Slide 29 Take the initiative today Summary Virtualization provides for: – Supporting specialized environments in a more resource effective manner – Migrating legacy platforms onto supported infrastructure – Implementation of a cost-effective, tiered Disaster Recovery solution – Opportunity to implement green computing while achieving ROI – Enhancing your agency’s security by moving computing from the desktop to the data center

U N C L A S S I F I E D LA-UR Questions? Slide 30