How VMware and VMware Capacity Planner Helped
Justify John Deere’s Transition to a Virtual Infrastructure
Bob DeBlaey, Infrastructure [email protected]
AgendaIntroductionDeere historyCurrent challengesOur history with VMwarePlanning our virtual infrastructureVMware Capacity Planner ChangesOur future
Our History1837 – John Deere develops the world’s first
commercially successful, self-scouring steel plow, opening the West to economic development
1918 – Modern tractor era begins (Waterloo Boy)1937 – Sales reach $100 million in centenary year1956 – Construction equipment business launched,
International manufacturing begins1958 – John Deere Credit Company launched1963 – Lawn and grounds care business launched1998 – Earnings reach $1 billion
John Deere TodayThe world’s leading producer of Agricultural EquipmentA Construction Equipment industry leader and world’s premier producer of timber-harvesting equipmentWorld leader in premium Commercial and Consumer turf-care equipment and work vehicles
John Deere Today
Mannheim, Germany
Richards Bay, South Africa
Nigel, South Africa
Pune, India
Jiamusi, China
Stadtlohn, GermanyGummersbach, Germany
Horst, NetherlandsEnschede, Netherlands
Bruchsal, GermanyZweibruecken, Germany
Arc-les-Gray, France
Saran, France
Madrid, SpainSantiago, Chile
Catalao, Brazil
Horizontina, Brazil
Rosario, Argentina
Saltillo, MexicoMonterrey, Mexico
Torreon, Mexico
Valley City, NDFargo, ND
Minneapolis, MN
Dubuque, IAHoricon, WI
Waterloo, IA
Moline, IL
East Moline, IL
Welland, Ontario
Edmonton, Alberta
Davenport, IA
Des Moines, IA
Langley, British Columbia
Ottumwa, IA
Springfield, MO
Coffeyville, KS
Thibodaux, LA
Fuquay-Varina, NC
Greeneville, TN
Augusta, GA
Kernersville, NC
Santo Angelo, Brazil
Joensuu, FinlandWoodstock, Ontario
Rock Valley, IA Klemme, IA
Tokoroa, New Zealand
Tianjin, China
Poznan, Poland
Orenburg, Russia
Agricultural Equipment
$9.7 Billion in SalesWorld’s premier farm machinery manufacturer
‘Ag Server Services’Transitioned from a unit based IT support group to a unified division based group
Our Mission: “Develop architectures, establish standards, define and implement services and support models that leverage IT resources to meet business goals at the highest possible value.”
“Virtual team” is distributed among, and supports, multiple locations across North America
Our Current Challenges
Containing server sprawl Server consolidation Out of warranty serversHigh Availability, Business Continuity, and Disaster Recovery planning
Our History With VMware
VMware Workstation: 2001 - 2004GSX Server: 2003 - 2005Only concern was scalability
The Solution
VMware ESX Server and VirtualCenter
The Plan
Perform a complete assessment of our current server environmentUse the data to begin our virtual infrastructure planning projectShow the financial benefits of a virtual infrastructure
The Plan
Show the hard dollar cost benefit and capital expenditures for each of our locationsProvide consolidation estimates Create a test ESX Server environmentDesign and document our architecture
Financial JustificationDeere uses Net Present Value (NPV) for evaluating the financial benefits of IT projects
NPV is the future stream of benefits and costs converted into equivalent values today. This is done by assigning monetary values to benefits and costs, discounting future benefits and costs using an appropriate discount rate, and subtracting the sum total of discounted costs from the sum total of discounted benefits
Our project estimated a 23.1% rate of return with a hurdle rate of 12%
Financial Justification
We estimated we could virtualize about 25%-50% of our total environmentPotential savings in future capital expenditures would save millions over the next five years
Additional AnalysisVMware Capacity Planner
Engaged VMware to provide us with some in-depth analysis and recommendations on our current physical server environmentThey recommended an assessment using VMware Capacity Planner
What Is VMware Capacity Planner and How Does it Work?
Uses native Windows Performance Monitor API’s to collect performance stats on “live”servers at specified intervalCollects data on a central server and uploads it to the Capacity Planner Dashboard Ran against our physical servers for the entire month of April 2005
Why Use VMware Capacity Planner?Validate our conclusions from our server analysis project Have enough “ammunition” to justify our actionsOptimize our investment in VMware software and hardwareProvides guidelines on capacity planning in a virtual world
VMWare Capacity PlannerImplementation Details
Required a physical serverLocal admin rights required on all target serversThe collector system must be authorized and able to connect to all the systems that it is to manage using the protocols and ports that are outlined in the VMWare Capacity Planner Getting Started GuideVerify that you are able to upload HTTPS through a proxy connection to the Capacity Planner site
Data Upload
Data Analyzer
HTTPS
InformationWarehouse
Infrastructure Assessments
Capacity Planning
Server Consolidation
Capacity Planner Dashboard
Discovery Inventory
Performance Data Synch
Data ManagerClient IT Environment
DataCollector
DataCollector
Agent-less Data Collection
ClientSite
HostedSecure Site
Web
Industry Data
VMware Capacity Planner: Implementation Details
VMware Capacity Planner Executive Summary
Data was analyzed, scrubbed, and presented to us by VMwareEstimated we could virtualize 80%-90% of our environmentConfirmed and surpassed all of our assumptions Provided us accurate, “real”, and useful data
VMware Capacity Planner ResultsServer Utilization Rates
Group
Available Memory
(MB) % CPUCPU
Queue Pages/secPageFile
% Av.Disk % Busy
Disk Av. Queue
Netw ork Traffic
2,453
1707.01 33,227
109,3795.38 2.34 0.31
0.81 1.62 0.07All Monitored 3.47 0.84 4.22
Industry Average 6.71 1.33 10.05
Well Below Industry Standards
VMware Capacity Planner ResultsNumber of Servers at nth% Utilization
Number of Servers at n% CPU Utilization
>0% <10% ≥10% <20% ≥20% <30% ≥30% <40% ≥40% <50% ≥50% <60% ≥60% <70% ≥70% <80% ≥80% <90% ≥90%≤100%
n% CPU Utilization
Num
ber o
f Ser
vers
Prime Time
Peak
VMware Capacity Planner ResultsActual Statistics
% Used Prime Time
% Used Peak Time
% Busy DiskQueue
W IDGETS SERVER1 2 0.190721636 0.600153502 1024 375.406276 1.6584596 0.016723799 0 37625.65153W IDGETS SERVER2 2 0.282436606 0.580775977 4096 927.245018 0.5891182 0.011782563 0 239057.783W IDGETS SERVER3 2 0.286981722 0.368084517 1024 437.284874 0.5934382 0.00593443 0.006510687 280917.7515W IDGETS SERVER4 2 0.348920168 0.542080322 512 234.865032 0.2749127 0.005499158 0.007049024 110926.1331W IDGETS SERVER5 2 0.493727623 0.737068778 2048 537.776464 0.3802531 0.007605009 0 12728.15459W IDGETS SERVER6 2 0.513437578 0.97715804 2560 384.524843 0.4614689 0.009229152 0.041778298 13326.6596W IDGETS SERVER7 2 0.554166466 1.83732027 2048 477.518246 0.4640688 0.009280126 0.031817223 31090.40903W IDGETS SERVER8 2 0.707956471 0.869523558 4096 1422.30202 0.113051 0.003390742 0.007371371 28820.39569W IDGETS SERVER9 2 0.713763089 2.00459124 2048 479.007855 0.7130034 0.007130209 0.000924659 33999.52134W IDGETS SERVER10 2 0.780703487 1.374716565 2304 505.553587 0.8827533 0.017655512 0.00397332 36325.55397W IDGETS SERVER11 2 0.825222679 1.440449039 2048 651.063895 1.6423018 0.03284587 0.073715049 43779.16431W IDGETS SERVER12 2 0.856825141 1.462333168 4096 1089.51497 0.387409 0.011622281 0.318331333 47832.25819W IDGETS SERVER13 2 0.937988998 3.208315836 4352 1278.3933 0.9261022 0.018521808 0 35571.53317W IDGETS SERVER14 2 1.068260246 1.632577938 1536 1165.02025 0.2266157 0.009064266 2.426333918 11161.82097W IDGETS SERVER15 2 1.109039185 1.551966033 4096 1140.81462 0.5404998 0.016213905 0.005438325 11197.13733W IDGETS SERVER16 2 1.150092206 1.303753206 4096 885.136445 0.5383119 0.016148925 0.206158546 11286.36809W IDGETS SERVER17 2 1.267322262 2.547243767 4608 1505.30562 1.0490728 0.031470931 0.167408822 11791.96767W IDGETS SERVER18 2 1.297821583 2.890420975 1536 555.275137 0.8960576 0.017924154 0.038604222 496681.2579
CPU Count
HostnameDomain
Performance Statistics
CPU UtilizationRAM Used
(MB)
Total RAM (MB)
Disk UtilizationPages/Per
Second
Network Bytes Per
Second
VMware Capacity Planner ResultsBad Candidates
Domain Hostname CPU Count CPU Utilization
Tota l RAM (MB)
RAM Used
Disk % Busy
Disk Queue
Netw ork Usage
W IDGETS BADSERVER1 2 43.7568025 1536 691.30662 0.502542 0.01508 18723.41W IDGETS BADSERVER2 2 52.8391443 1536 605.65714 3.264448 0.09793 19204.91W IDGETS BADSERVER3 2 58.2158906 1536 658.62053 12.21732 0.31498 16519.64W IDGETS BADSERVER4 2 67.2799959 4608 1916.7213 19.32431 0.59732 11915.11W IDGETS BADSERVER5 2 71.7332187 4096 680.10761 4.866089 0.15453 87472.41W IDGETS BADSERVER6 2 76.276942 2048 684.26865 1.423794 0.08742 17823.15W IDGETS BADSERVER7 2 85.8844797 2304 814.28981 19.28764 0.44894 24927.63W IDGETS BADSERVER8 4 24.2875611 4096 1126.8722 22.71291 0.60632 46912.71W IDGETS BADSERVER9 4 23.6242964 4096 1118.731 17.59259 0.38299 47365.28
Bad Candidates for Virtualization
Any System utiliz ing more than 40% of 2 CPUs and/or 3.6GB of memory is determined to be a bad candidate for Marked in RED is the resource utilization that indicates this workload may not be suitible for running in a virtual
VMware Capacity Planner ResultsReusable Servers
Domain Hostname Chassis Model CPU Count
CPU Speed
RAM (MB)
Disk Space (GB) NIC(s)
WIDGETS GOODSERVER1 ProLiant DL380 G3 2 2791 4096 1230.58823 7WIDGETS GOODSERVER2 ProLiant DL380 G3 2 2791 4096 1230.58823 8WIDGETS GOODSERVER3 ProLiant DL380 G3 2 2791 2048 820.392151 7WIDGETS GOODSERVER4 ProLiant DL380 G3 2 2791 4096 820.392151 7WIDGETS GOODSERVER5 ProLiant DL380 G3 2 2791 4096 820.392151 7WIDGETS GOODSERVER6 ProLiant DL380 G3 2 2790 2048 410.196075 7WIDGETS GOODSERVER7 ProLiant DL380 G3 2 2790 2048 820.392151 8WIDGETS GOODSERVER8 ProLiant DL380 G3 2 2790 2560 820.392151 7WIDGETS GOODSERVER9 ProLiant DL380 G3 2 2790 2048 820.392151 8WIDGETS GOODSERVER10 ProLiant DL380 G3 2 2790 2048 820.392151 7WIDGETS GOODSERVER11 ProLiant DL380 G3 2 2789 4608 1230.58823 6WIDGETS GOODSERVER12 ProLiant DL380 G3 2 2789 2560 1230.58823 8WIDGETS GOODSERVER13 ProLiant DL380 G3 2 2789 4096 313.25592 7WIDGETS GOODSERVER14 ProLiant DL380 G3 2 2788 4608 1230.58823 6WIDGETS GOODSERVER15 ProLiant DL380 G3 2 2788 4096 1230.58823 7WIDGETS GOODSERVER16 ProLiant DL380 G3 2 2786 4096 1230.58823 8WIDGETS GOODSERVER17 ProLiant DL380 G3 2 2786 4608 1230.58823 8WIDGETS GOODSERVER18 ProLiant DL380 G3 2 2785 4096 1230.58823 7WIDGETS GOODSERVER19 ProLiant DL380 G3 2 2784 2048 104.41864 7WIDGETS GOODSERVER20 ProLiant DL380 G3 2 2784 4096 820.392151 7WIDGETS GOODSERVER21 ProLiant DL380 G3 2 2784 4096 156.62796 8WIDGETS GOODSERVER22 ProLiant DL380 G3 2 2783 4096 1230.58823 8WIDGETS GOODSERVER23 ProLiant DL380 G3 2 2392 2560 820.392151 7WIDGETS GOODSERVER24 ProLiant DL380 G3 2 2388 4096 156.62796 7WIDGETS GOODSERVER25 ProLiant DL380 G3 2 2386 2560 410.196075 7
Repurpose Servers as Virtualization Platform ServersHere is the list of all systems in the All Servers (1, 2, 4 CPU) consolidation scenario that potentially could be repurposed as Virtualization Platform Servers or for other purposes.
VMware Capacity Planner ResultsObsolete Servers
Domain Chassis Model CPU Count
CPU Speed
RAM (MB)
Disk Space (GB) NICs
WIDGETS SERVER1 NetServer LH 3 1 399 512 50.84966183 5WIDGETS SERVER2 NetServer LH 3 1 399 512 33.8895607 5WIDGETS SERVER3 NetServer LH 3 1 399 512 33.89722109 5WIDGETS SERVER4 NetServer LH 3 1 399 1024 33.89722109 5WIDGETS SERVER5 NetServer LH 3 1 399 768 33.90488148 7WIDGETS SERVER6 NetServer LH 3 2 399 1024 59.322052 5WIDGETS SERVER7 PowerEdge 6300/400 4 400 4096 3.968081474 5WIDGETS SERVER8 PowerEdge 6300/400 4 400 4096 16.94478035 5WIDGETS SERVER9 HP NetServer 4 550 4096 152.6102686 5WIDGETS SERVER10 HP NetServer 4 550 4096 152.6102686 5WIDGETS SERVER11 HP NetServer 4 550 4096 152.6102686 5
Obsolete ServersThe following table lists all obsolete servers and their specifications. All systems listed here in RED candidates for virtualization in the All Servers (1, 2, 4CPU) consolidation scenario.
Project ResultsCurrently 25% of our environment virtualized since AugustNearly 50% of environment will be virtualized by end of yearCurrent CPU utilization on ESX hosts under 15%Memory utilization under 15%Consolidation ratios could easily hit 20:1
Culture Change
All servers will now be virtualized Server builds take minutes Response time to business is much quickerOther units and divisions are following our lead
Process ChangeDisaster Recovery and Business Continuancyplans have been greatly simplified Vmotion has enabled us to become more flexibleVirtual Machines specs based on Capacity Planner resultsOur capacity planning analysis and skills must change Technical skills have been updated
Our Future
Continue Consolidating ServersDual-core serversVirtual machine clustersContinue integrating IT groups from other divisionsOptimize our current environment
Questions