CA Capacity Management Community Webcast Automated Modeling Methods
CA Capacity Management Community Webcast Automated Modeling Methods Service Forecasting Kip Lamb Sr Principal Consultant Jeremy Rossbach Sr Product Marketing Manager January 22, 2015
Today’s conversation § Preview of upcoming CA Cap. Man events § Customer Problems Today § Examples & Opportunities § Proposed Approach to Model Automation § Demo 2 © 2014 CA. ALL RIGHTS RESERVED.
Next Week: SPRINT tells their CA Capacity Management Success Story – Tues, Jan 27 th, 1 pm EST – Register here: http: //bit. ly/1 Afq. Ja. M – Or check out the recent postings in the CA Capacity Management Community 3 © 2014 CA. ALL RIGHTS RESERVED.
Next Week: Monthly CA Capacity Management Demo Series – Thurs, Jan 29 th, 1 pm EST – Register here: http: //bit. ly/1 Ds. C 4 d. R – Or check out the recent postings in the CA Capacity Management Community 4 © 2014 CA. ALL RIGHTS RESERVED.
Abstract Automated Capacity Modeling Due to limited resources, most IT organizations only plan for capacity on a subset of resources supporting critical applications. As a result, optimization for lower tiered applications and infrastructure usually does not happen. In this lab, you will be exposed to a new automated capacity modeling and projection methodology based on CA Capacity Management that can ensure ALL your applications are monitored and modeled on a continual and automated basis. 5 © 2014 CA. ALL RIGHTS RESERVED.
Four Universal Questions of Capacity Planning These questions we answered 30 years ago are the same we strive to answer today. § What is the problem? § How bad is the problem? § How long before I need to resolve the problem? § What do I need to do to mitigate the problem? 6 © 2014 CA. ALL RIGHTS RESERVED.
Customer Challenges Today § 80/20 Rule Applies – due to lack of time/resources customers prioritize subsets of systems to “model” which is often limited to tier 1, customer facing applications or those responsible for revenue generation. § As a result of prioritization……optimization most often does not happen. § Repetition – there is most often little to no variation in models that are run from month to month. § Capacity Planners do not have credentials for most systems they plan for…. they are just the messenger. Priority of an OS instance does not mean it costs less. 7 © 2014 CA. ALL RIGHTS RESERVED.
Capacity Management Maturity Model Most customers are here 8 © 2014 CA. ALL RIGHTS RESERVED.
CA Capacity Management Process Steps Most steps don’t require human intervention Collect and analyze data 9 Identify potential risk points Model Whatif to resolve potential bottlenecks © 2014 CA. ALL RIGHTS RESERVED. Provide Results that Reduce Risk
Circle of life…………. for a Capacity Planner 1. Configure Servers 7. Evaluate Rinse & Repeat Repetition Presents Opportunities § – Automation enables Scale 6. Apply Change – Automation enables Options – Automation allows for continual Optimization Automate 3. Set Groups/Tiers 5. Select Baseline 4. Select Workloads 10 © 2014 CA. ALL RIGHTS RESERVED. 2. Set Thresholds
CA Capacity Management Modeling Scenarios Top 2 scenarios represent the most commonly performed (70%)* § Workload Changes (40%) § Configuration Changes (30%) § Virtualization (15%) § Server Consolidation (10%) § Workload Reassignment (5%) * Source: CA Sponsored survey 2014 11 © 2014 CA. ALL RIGHTS RESERVED.
Automated Modeling Features We will cover current features today. What would you like to see added? § All servers – any metric § – Configurable & determined by user § Customizable Historical Intervals – Calculated by system & interval § – Allow for multi-baseline comparisons § Automated Data Integrity Analysis Automated Historical Regression Calculation – Includes slope and intercept Multi-baseline computation options § Configurable Projection Scenarios – By Group/Service – Avg, 90 th Ptile, 95 th Ptile, MAX 12 © 2014 CA. ALL RIGHTS RESERVED. For CPU, the Rx Utilization is used
Resource Score What is it? Why does it matter?
Capacity (Rx) Utilization vs. ‘Regular’ Utilization Regular CPU utilization is not linear. But, Capacity Utilization is… 100% 75% 50% 25% 0% n 0 n 1 n 2 n 3 Capacity Utilization 14 n 5 Regular Utilization © 2014 CA. ALL RIGHTS RESERVED. n 6 n 7 n 8
Automated Capacity Modeling – Projections using Rx Scores § Resource Scoring – Projections/results are linear so simulations are not bound to a specific tool. – Current State (Capacity) – 55. 7% – 55. 07 x 1. 50 = 82. 605% – 55. 07 x 1. 75 = 96. 3725% 15 © 2014 CA. ALL RIGHTS RESERVED.
CA Modeling Technology 16 HW Inefficiency VMM Overhead Reserve Headroom Capacity Manager Model Guest OS 2 Overhead VM 2 Raw Consumption Move Guest OS 1 Overhead VM 1 Raw Consumption © 2014 CA. ALL RIGHTS RESERVED. Move Add Old System Remove Capacity Manager models can factor out the overheads introduced by hardware, OS’s and virtualization hypervisors allowing fast and accurate what-if of moving workloads (in terms of Raw Consumption) to a different server platform New System Headroom Guest OS 2 Overhead VM 2 Raw Consumption Guest OS 1 Overhead VM 1 Raw Consumption
Service Profile Reporting Service Profile Summary Landing Page § Historical Analysis (Service Profile) – Past § Current State (Service Optimization) - Present § Future Projections (Service Forecast) – Future 17 © 2014 CA. ALL RIGHTS RESERVED.
Historical Analysis (Service Profile) – Past § Histograms – Usage & Distribution – Threshold Distribution § Days to Threshold – Allow for multi-baseline comparisons – Data Integrity § Baseline Calculation Details – Avg, 90 th Ptile, 95 th Ptile, MAX 18 © 2014 CA. ALL RIGHTS RESERVED.
Current State (Service Optimization) - Present § Service Optimization Summary – Total Reclaimable Assets § Independent Metric Analysis – Normalized (Rx Score) consumption values – Evaluation of current state and regression (trend) of assets 19 © 2014 CA. ALL RIGHTS RESERVED.
Projections (Service Forecast) – Future § Multi-baseline Comparison § Configurable Scenarios – Growth & Projection Intervals – By Group/Service *not included in beta* § Customizable Metric Analysis – Includes Dynamic Memory 20 © 2014 CA. ALL RIGHTS RESERVED.
Automated Modeling Features 3 Simple Steps to Setup Forecast 21 © 2014 CA. ALL RIGHTS RESERVED.
Demo
Q & A 23 © 2014 CA. ALL RIGHTS RESERVED.
- Slides: 23