IT Economics: A Structured Approach to Identify, Measure and Reduce IT Costs David Merrill Chief Economist Hitachi Data Systems
Leverage IT Economics
Econ 101 Demand is growing Budgets are flat or declining How to reconcile? Year 1 Year 2 Year 3 Year 4 Year 5 Address and reduce UNIT COST
Common IT Challenges Shrinking budgets Organizational change Increased competition Regulation, compliance and data management Power, cooling and limited floor space Increased cost of managing data VM Sprawl, affordable plans for big data growth Inefficient (or old) infrastructure architectures Confusion in cloud pricing, TCO, benefits
Economic Concepts, principles It is becoming increasingly essential to apply economic and financial principles to IT Architectures, roadmaps, standards Operational excellence Consumption behaviors Use TCO to measure / compare Identify Measure Reduce ROI and ROA to cost justify Measure Again 4 key principles of storage and IT economics 1. Price does not equal cost price is about 20% of TCO 2. 34 different types of cost where is your sensitivity 3. There are economically superior IT architectures 4. Econometrics You cannot improve what you cannot measure
A STRUCTURED APPROACH Identify Choose from the 34 types of cost that are relevant Get CxO and technical agreements on cost areas Apply this process to storage, VM, VDI, cloud infrastructure Measure Calculate and measure the unit cost baseline By tier, by platform and by geography Isolate direct and indirect costs & determine cost owners Reduce Map costs to investments that reduce costs Predict unit cost reduction or ROI Set plans for transformation Measure results frequently
Driving down unit costs of IT Phase 1: The Basics Phase 2: Virtualization Phase 3: New Behaviors Phase 4: New Owners, Locale
TCO Baselines - start your econ journey Decide a start-point (before a major transformation) Select the costs that are important (from the 34) No right or wrong approach, but most choose 12-15 categories Differentiate direct and indirect To start, we need 4 basic parameters Total usable capacity Average age of storage assets Recent growth rate Country / region DC floor sapce Power, Cooling SLA penalties: Backups Scheduled Downtime Outage Risk Contract Labor Internal Labor HW Maintenance SW Maintenance Purchase ~50 more detailed questions can provide better precision Baseline then moves to the mapping phase Baseline TCO
Mapping Costs to Storage Solutions 1. Hardware depreciation (lease) 2. Software purchase or depreciation 3. Hardware Maintenance 4. Software Maintenance 5. Storage Management, Labor 6. Backup and DR Labor 7. Migration, Remastering 8. Data mobility 9. Power Consumption/Cooling 10. Monitoring 11. Data Center Floor space 12. Provisioning Time 13. Cost of waste 14. Cost of copies 15. Cost of duplicate data 16. Cost of growth 17. Cost of scheduled outage 18. Cost of unscheduled outage (machine) 19. Cost of unscheduled outage (people & process) 20. Cost of disaster risk, business resumption 21. Recovery time (RTO) Costs 22. Data Loss 23. Litigation, discovery risk 24. Reduction of hazardous waste 25. Cost of Performance 26. Backup infrastructure 27. Backup media 28. Cost of risk with backup windows 29. CIFS, NFS related infrastructure 30. Local and remote data circuits 31. Storage area networking 32. Non-compliance risk (archive, data retention) 33. Security, encryption 34. Time for procurement A. Virtualization B. Thin Provisioning C. Dynamic Tiered Storage D. Storage Intermix E. Active Archive F. De-duplication G. SSD H. Zero Page Reclaim I. Storage Consolidation J. SAN Consolidation K. ITIL Best Practices L. Storage Team Organization M. Disaster Recovery N. VTL O. Disk Based backup P. Storage Area Network Q. Unified Backup R. Unified Console, Advanced Mgmt S. Policy based management T. Storage Architecture U. Storage Services Catalog V. Utility Services W. Outsourcing X. Managed Services
TCO/TB/Year (thousands) Case Study Focusing on Unit Costs 12.00 10.00 8.00 6.00 4.00 2.00 0.00 Blended Tiers TCO/TB/Year 2007 2008 2009 2010 2011P 2011V Cost of Disaster Risk DR Infrastructure SRM (Tools, labor) Floor space Power, cooling Data migration, remastering Storage Management, Labor Software Maintenance SAN HW Maintenance Storage HW Maintenance Software purchase or depreciation Storage File Servers Storage circuits (DR, dark fibre) SAN HW depreciation Storage Waste Storage HW Depreciation
Key Investments to Reduce Unit Costs Structured, funded investment plan DR Facility, Test NAS Consol Tier 1 Virtualize Tiering SAN Consol De-dupe Thin Provision Tier 2 Virtualize
TCO/TB/Year (thousands) Unit Cost Reductions- 7 Year Trend 12.00 10.00 Adobe Blended Tiers TCO/TB/Year Global Capacity Cost of Disaster Risk DR Infrastructure SRM (Tools, labor) Floor space Power, cooling 8.00 6.00 4.00 2.00 Data migration, remastering Storage Management, Labor Software Maintenance SAN HW Maintenance Storage HW Maintenance Software purchase or depreciation Storage File Servers Storage circuits (DR, dark fibre) 0.00 2007 2008 2009 2010 2011V 2012 2013 2014 SAN HW depreciation Storage Waste Storage HW Depreciation 2014 Blended unit cost rate is $0.80/GB/Year Note TCO measurements not taken in 2010, 2012 and 2013 Average of 31% Y/Y unit cost reduction for the blended rate of all tiers
Hypervisor Economics MEASURING AND REDUCING THE TOTAL COST OF A VIRTUAL MACHINE 13
Common Problem Areas Pain in scaling Time to provision Overall growing costs especially license costs Threat of public cloud offerings Coordinated management Moving RISC workload to Linux on Intel Consolidation of storage resources Governance of the converged estate Slow down VM Sprawl Reduce VM growth costs Simplified management reduced labor time, effort and eventually FTE
TCO per VM/Month VM and CI Cost Evolution Time-based View of VM Costs 1,000 mixed workload VM, US on-premise DIY VM 1st Gen CI 2nd Gen CI Orchestration
What goes into the VM Costs 1. Hardware depreciation expense 2. Software purchase or depreciation 3. Hardware Maintenance 4. Software Maintenance 5. General Admin/ Management, Labor 6. Migration, Remastering 7. Data or workload mobility 8. Power Consumption & Cooling 9. Monitoring 10.Data Center Floor space 11.Provisioning Time 12.Cost of waste (or reserve capacity) 13.Cost of growth 14.Cost of scheduled outage 15.Cost of unscheduled outage (machines) 16.Cost of unscheduled outage (people/ process) 17.Cost of disaster risk, business resumption 18.Cost of Performance 19.Backup infrastructure 20.CIFS, NFS related infrastructure 21.Local &remote data circuits, equipment, including SAN and NAS 22.Security, encryption 23.Cost of Procurement 24.Integration, Test and System Certification
Some ideas, Options to Reduce Cost 1. Tech refresh the server estate, standard systems config 2. Refresh the storage estate 3. Higher density servers reduce servers, racks, environmentals (higher licensing?) 4. Physical to Virtual for the next wave of systems (below 50%) 5. Upgrade or replace the Hypervisor 6. VASA, VAAI (vvols) 7. Tight integration of management tools, console, native VM tools 8. Recommend certified, tested SKU linked to ref arch 9. Create ref architecture, standard service catalog; limit customization (service and size matrix) 10. Limit the number and types of VM options available 11. Charge-back for the VM 12. Self provisioning portal for users or IT staff to select from the offering 13. Lifecycle management, including de-commission process 14. Move RISC to Intel servers, LPAR 15. Orchestration software, and end-to-end process alignment 16. Organizational changes for CI, integrated management & control (ITIL best practice) 17. Move away from depreciation/capex to elastic financial model (PPU, on-demand, utility) 18. Move to public utility or cloud 19. Hybrid cloud for bursting 20. Annual re-measure and reporting on unit costs 21. Cluster pairing for HA critical apps 22. 2 nd vendor for servers, storage 23. Move from DAS to SAN/NAS pooled storage 24. Software defined networks, data center (ease of use, transfer) 25. Lower tiered storage (archive, object store) 26. Archive to reduce the cost of backup 27. Better monitor, thresholds, trends to improve performance, uptime; health predictive
Case Studies on CI Economics Unit Cost Summary of Phys & Virt Server Physical servers Avg w/ucp & SAN Average VM weighted Windows Small Linux Small Windows Medium Linux Medium Windows Large Linux Large Linux Special Purpose Windows Special Purpose - 1,000 2,000 3,000 4,000 5,000 6,000 $ / VM / Year
Cloud Economics SHIFTING OR REDUCING COSTS WITH A CLOUD 19
Cloud Economics Common economic principles applied to clouds Price cost Different types of costs, and different cost locations Metrics are key to track your costs and 3rd-party cost Architectures differ as each definition of cloud differs Cloud offerings each present new cost options, some actually reduce, others simple transfer Capacity on demand MSS, MSU Private, hybrid or public IaaS options Converged solutions and pods Must differentiate cloud technology and cloud procurement options
Modeling cloud economics
Don t be seduced by Cloud Price TCO per VM/Month TCO per GB/Month
$/VM per Month Econ benefits of VM in Private Cloud VM TCO in customer own data center, local mgmt Fast provisioning. Self serve No patch mgmt, engineering, SLA mgmt UCP HW Benefits Charge-back, catalog, metering, cost allocation Other (life cycle mgmt, ELA, economies of scale, migration, flex down) VM TCO in a managed private cloud Some of the VM unit cost benefits comes directly from the UCP platform (storage, consolidation, mixed workload) Many of the costs comes from shared services and advanced operations in the cloud (provisioning time, chargeback, operations, skilled staff, flex up and down, economies of scale with Hitachi purchasing power) Even with the cloud subscription, the customer will still retain some other costs (network, contract mgmt, SLA mgmt)
Conclusion Hope is not a strategy in Data Center efficiencies and cost reduction Direct efforts, strategies are needed to reduce real costs over time Make sure you target the costs that are important, and choose investments (people, process, technology) that are proven to reduce these costs New measurements, metrics, incentives and behaviors are needed to have an impact on data center total costs A mix of technology, best practices and consumption behavior change will be needed to reduce real costs Piece-meal or holistic approach your choice Current IT growth trends are economically unsustainable, track where you are and make plans for a new unit cost trajectory