Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Capacity and Demand Management

20,843 views

Published on

Service based / modeled IT operations demands that Infrastructure needs are catered to with minimal disruptions and loss of user experience. Demand and capacity management for a critical cog in IT / service design to ensure that the service / infrastructure is fully available to users through its lifecycle

Published in: Technology
  • Well done!
    Great job in simplifying a very complex subject.
    Enjoyed reading it and love the story flow.
       Reply 
    Are you sure you want to  Yes  No
    Your message goes here

Capacity and Demand Management

  1. 1. Demand & Capacity Management Vishwanath Ramdas
  2. 2. Capacity – Demand Management framework Identify usage patterns both short term and long term translate customer demands into workloads put upon critical components Monitor and collect performance data of critical components in current configuration Determine the current and future resource needs. Ensure resources are acquired & implemented in a timely and cost-effective manner Demand Management Performance Management Workload Management Resource Management
  3. 3. What are the criticality Maps to setup monitoring? Demand Management .a 1 Define the Service Catalog within IT Develop the services catalog. <ul><li>IT services catalog is generated across all functions within IT </li></ul><ul><li>The Team sits with each function and lists down the services provided by each function </li></ul><ul><li>Services are defined from an outside in perspective [ business users perspective] </li></ul><ul><li>Services would be </li></ul><ul><ul><li>Business Applications </li></ul></ul><ul><ul><li>Internal Applications </li></ul></ul><ul><ul><li>Infrastructure Applications </li></ul></ul><ul><ul><li>IT Infrastructure Services </li></ul></ul>The service catalog is an overall IT asset and would be helpful in IT SLM process.
  4. 4. What are the criticality Maps to setup monitoring? Demand Management .b 1 Define parameters that would help classify services. Develop the services catalog. <ul><li>Parameters along 2 dimensions </li></ul><ul><ul><li>Critical to Business </li></ul></ul><ul><ul><li>Critical on Capacity </li></ul></ul><ul><li>WIPRO would bring a ready template to define parameters. </li></ul><ul><li>Finalize these key parameters for classifying IT services with IT managers in a brainstorming session. </li></ul><ul><li>Define the service catalog data collection </li></ul><ul><li>Create the service classification template of the service catalog. </li></ul>The service catalog is an overall IT asset and certain data in the service catalog would be useful for capacity – demand planning and management.
  5. 5. What are the criticality Maps to setup monitoring? Demand Management .b 1 Service Catalog developed for a petrochem - MNC Sample of a developed services catalog as part of SLM workshop. The Service catalog workshop defined the parameters of each service and listed all key IT services and created an SLA for each service
  6. 6. What are the criticality Maps to setup monitoring? Demand Management .c 1 <ul><li>While listing the services with each function, the capacity parameters are collected based on past experience, issues and data </li></ul><ul><ul><li>Typically capacity criteria are picked from past 6 Mo performance / legacy </li></ul></ul><ul><li>Finally for the services identified, the Business parameters are defined by a central team with visibility into the business </li></ul><ul><li>The Service catalog listing could either be in a database or in a spreadsheet format. </li></ul>Get Ratings from IT Managers on the business and capacity criteria. Classify and define capacity management approaches
  7. 7. What are the criticality Maps to setup monitoring? Demand Management .d 1 <ul><li>Services in 1st Quadrant are both business and capacity critical & would require Active Monitoring. </li></ul><ul><li>Services in 2nd Quadrant are Capacity Critical but not business critical, use Resource Monitoring </li></ul><ul><li>Services in 3rd Quadrant are only Business critical and they could be managed with periodic reviews and long term forecasts </li></ul><ul><li>Services in 4th Quadrant are neither critical and these are good targets for consolidation. </li></ul>Visualize the services map & select capacity management strategy based on classification. Classify and define capacity management approaches 2 1 4 3
  8. 8. Mapping Business Drivers to Infrastructure Capacity <ul><li>For selected Application services, define the Main Use Cases. </li></ul><ul><li>For each Use Case, identify what are the infrastructure resources required by defining a Component Capacity Impact Matrix. – QFD Format </li></ul><ul><li>The CCIM requires current monitoring data or heuristics to define the quantum of influence </li></ul>Workload Management 2 Define the Use Scenarios for each service and build the Component Capacity Impact Matrix [CCIM] .a Applicable for 1 st Quadrant Services / Active Monitoring. – Mapping demand to capacity. The CCIM is a critical asset to IT, it can control and provide data on business value of IT if Value metrics are drawn for the services. This can then trace the value of each infrastructure element back to business.
  9. 9. Mapping Business Drivers to Infrastructure Capacity 2 Identify Business drivers that influence capacity. <ul><li>For each service use case identify Business drivers in terms of users volume, Concurrency, Revenue, Transaction volume, data size. </li></ul><ul><li>This is contextual to each service and the business context. </li></ul><ul><li>Map the business drivers to infrastructure element using the Capacity Impact Matrix. </li></ul>.b Applicable for 1 st Quadrant Services / Active Monitoring. – Mapping demand to capacity. Supply side Demand side Workload Management
  10. 10. Modeling Long term Aggregate Demand Forecasting 3 Define the Capacity Planning process .a Create a demand forecasting & capacity planning framework Defining the capacity thresholds for a exchange service for a large Indian MNC <ul><li>SLM Planning process provides inputs on new services being introduced into the system. </li></ul><ul><li>Regular review of the above activities and models by capacity planning </li></ul><ul><li>Capacity planning does an impact analysis of new services on common infrastructure CI. </li></ul><ul><li>Capacity planning should also ensure execution of changes in infrastructure based on analysis - roadmap </li></ul><ul><ul><li>What and how of change </li></ul></ul><ul><ul><li>Budget process </li></ul></ul>Performance Management
  11. 11. Modeling Long term Aggregate Demand Forecasting 3 <ul><li>Build the Model [spreadsheet / Modeler software] </li></ul><ul><ul><li>Would be refined regularly through monitoring. </li></ul></ul><ul><ul><li>A software would provide simulation and provide scenario analysis and what if analysis. </li></ul></ul>For a education services leader In UK, their capacity requirements were simulated and finally modeled in EXCEL Determined the network bandwidth for a large Digital Signage SAAS service with potential 100K screens. Create a demand forecasting & capacity planning framework Build the long term Demand & capacity modeling framework .b .1 .2 Performance Management
  12. 12. Samples of Long term aggregate demand Forecasting 3 Capacity – Demand Forecasting for Hosted services for a European Education Services Co 2 3 <ul><li>Performance Benchmarks were conducted on the systems on simulated load and data from 70% capacity utilization taken </li></ul><ul><li>Key performance metrics at std utilization level were captured for all critical software elements across application and database </li></ul><ul><li>Based on the measurement data a forecasting model was built to estimate what would be the capacity requirements at different load points. </li></ul>.1 Performance Management
  13. 13. <ul><li>Defined the key services provided by the ISP. </li></ul><ul><li>Translated the services into transactions requests within the system </li></ul><ul><li>Heuristic / empirical translation of the transaction requests as workload on infrastructure </li></ul><ul><li>Modeled the infrastructure on a queuing model using process simulation tool </li></ul><ul><li>Modeled the infrastructure for various demand scenarios using demand distributions across time. </li></ul><ul><li>Developed infrastructure specifications and </li></ul><ul><li>Thereby developed infrastructure setup plan. </li></ul>Samples of Long term aggregate demand Forecasting 3 Capacity – Demand Forecasting for infrastructure capacity of hosted service provider .2 Performance Management
  14. 14. Different Monitoring approaches for critical applications System monitoring, setup measurement agents for capturing, CPU, memory, storage space, process metrics and OS metrics on the server & network nodes. .a .b 4 Different ways to monitor infrastructure. End user end to end transaction monitoring. This is done through remote agents that create synthetic transactions and trap various response times for each use case transaction. End User Monitoring was setup for a large IT services / consulting organization for their central – web project management software Performance Management
  15. 15. Different Monitoring approaches for critical applications Performance Management Capture logs and traces from web servers and DB servers to capture both response times for specific transactions and the patterns of usage of the resources. What are end users demanding? Transaction tracing within application servers to capture execution of software code on systems. Used in software performance engineering in load simulated environments. Used with prior care in production environments as these monitors can consume significant resources. .c 4 .d Different ways to monitor infrastructure.
  16. 16. Samples of Monitoring setup for different clients. Resource Management 4 Typical capacity monitoring done for a large US MNC. Trend view of hosting center bandwidth consumption from AT&T services. These trends are used as inputs for fine tuning the demand models and for capacity planning. Capacity usage trends for staff usage for the network service, showing available capacity and usage. Such a trend was used to re-contract and redeploy resources in network management and support
  17. 17. Modeling Capacity (Six Sigma: Process Capability) Identify critical apps managed by the DC, define performance criteria and setup the monitoring systems Map and identify demand scenarios in critical apps [ in pilot – 2/3 apps] – monitor demand - sys monitors Correlate application into infrastructure elements and collect data on infrastructure Join demand & infrastructure performance and monitor capacity metrics w.r.t limits App - DC inventory 1 2 4
  18. 18. Forecasting (Six Sigma: Process Trends) Analyze the performance across parameters Forecasting using statistical tools like trend analysis, seasonality, c/r-charts <ul><li>7 patterns analysis of C – Charts </li></ul><ul><li>Trends with seasonality settings to ensure peak load </li></ul><ul><li>Demand MACD [moving average convergence divergence] curves which are moving average exponential 2 sigma UCL LCL to show how demand is </li></ul><ul><li>Bottleneck analysis across infrastructure streams for each application / service. </li></ul><ul><li>Across time, infrastructure elements </li></ul><ul><li>Across all elements of a transaction to see where constraints lie. </li></ul><ul><li>Define the throughput at each stage and find the constraining elements </li></ul>
  19. 19. Setting the Limits ( Six Sigma : Process Control ) <ul><li>Define relationship between the demands and workload, </li></ul><ul><li>Monitor resource workloads and utilizations, so that appropriate thresholds can be set at each level. </li></ul><ul><li>Analysis follows with reports, & recommendations </li></ul>Interpret the trends - event management and alerts. Setup Capacity governance and control procedures based on ITIL standards and 6sigma control <ul><li>Event Collection & Logging </li></ul><ul><li>Event Consolidation: Filtering & Classification </li></ul><ul><li>Event Correlation & Enrichment </li></ul><ul><li>Event Action / Processing </li></ul><ul><li>Event Closure </li></ul>
  20. 20. What are solutions to identified constraints? 3 approaches are presented here based on lean tenets. <ul><li>Capacity planning and monitoring help identify Wastes [Under Utilized Resources] </li></ul><ul><ul><ul><li>Consolidate </li></ul></ul></ul><ul><ul><ul><li>Retire </li></ul></ul></ul><ul><li>These wasted resources are the Quadrant 3 & 4 resources in the services criticality map. </li></ul>Find the bottleneck and remove the non value add wastes in the system [MUDA] 1
  21. 21. What are solutions to identified constraints? 3 approaches are presented here based on lean tenets. <ul><li>Weakest link in the chain reflects the strength of the chain. </li></ul><ul><li>Capacity planning and monitoring help identify Bottlenecks [Over utilized Resources] </li></ul><ul><ul><ul><li>Optimize the application </li></ul></ul></ul><ul><ul><ul><li>Increase the capacity to meet demand </li></ul></ul></ul>Balance demand & Capacity [MURA]. Level demand across time and make capacity flexible to varying demand 2
  22. 22. What are solutions to identified constraints? 3 approaches are presented here based on lean tenets. <ul><li>Capacity planning and monitoring help identify overload in the system </li></ul><ul><li>Excess data flow in the system – network? </li></ul><ul><ul><li>AJAX | JSON </li></ul></ul><ul><li>Sub optimal code execution </li></ul><ul><li>Service Oriented Architecture. </li></ul>Innovatively expand capacity or shrink processing to obviate overload [MURI] 3
  23. 23. Vishwanath Ramdas is … <ul><li>Business Process Operations Consultant </li></ul><ul><li>Working in areas like ITSM, Software development, process consulting … </li></ul><ul><li>Based in bangalore [India] </li></ul><ul><li>Visit him @ </li></ul><ul><ul><li>http://gopu44.googlepages.com </li></ul></ul>Thank You

×