VMware vCenter Operations Standard Real-time Performance Management for VMware Administrators Technical Presentation © 2010 VMware Inc. All rights reserved Why vCenter Operations Standard? 80% of VMware admin time spent isolating performance problems • “1st generation” green-yellow-red static threshold reporting insufficient and too complex to use • Point solutions only address a subset of issues VMware administrators have two conflicting goals • Maximize ROI by increasing VM density • Ensure required capacity for business growth and other changes in real-time • Ensure that virtual component performance supports required application performance 2 VMware vCenter Operations Standard Basics Clear and quick way to identify VMware performance problems Easy to use for VMware Administrators • Deeply integrated as a vCenter pane • Intuitive screens guide users to issues needing attention • Automatically collects data from vCenter • Time-series performance data, topological relationships and configuration change events VMware vCenter Operations Standard business benefits • Increased performance for end users of business applications and services • Reduced infrastructure costs through increased VM to ESX density • Reduced VM administration costs and optimized VMware admin productivity 3 VMware Named 2011 Best of Interop Grand Prize Winner vCenter Operations Standard 1.0 named Best of Interop 2011 out of 135 award entries vCenter Operations Standard 1.0 honored as the winner in Cloud Computing and Virtualization What the judges said… • 16 editors and analysts from InformationWeek Analytics “VMware's vCenter Operations is another one of those products where you can see an ambitious company testing the limits… It is a bold effort to combine the data center disciplines of system configuration, performance management, and capacity management into one management tool and apply them to what in the future will be referred to as the private cloud.” Charlie Babcock, Editor at Large "VMware and each of the category winners represent market innovation and deserve recognition for helping build the energy and growth of today's IT marketplace." Art Wittmann, Vice President & Director 4 "We are pleased to have Interop be the platform for bringing together the latest advancements in business technology and we congratulate VMware for their innovation in the industry.” Lenny Heymann, Interop General Manager Understanding your Virtual Environment - Workload Workload Measures • Demand for resources vs. Resources currently used • Result is a percentage of Workload • Low number is Good – Object has the resources it needs • Can go above 100% - Object is “Starving” Workload summarized across critical resources • CPU • Network I/O • Storage I/O • Memory (VM and ESX Allocation) Workload Details View • Detailed understanding of the lacking resource and associated metrics • View the state of the Peer and Parent Objects and troubleshoot • Am I a victim or a villain? • A Configuration issue? • Is this a population problem? • Lack of resources? • Should we move the VM? • Virtual infrastructure is fine. OS or application issue? 5 Understanding your Virtual Environment - Health Health Measures • How normal is this object behaving: 0-100 (Higher is Healthier) • Learns dynamic ranges of “Normal” for each metric • Learns patterns of behavior and identifies metric abnormalities • Lower the health the more abnormalities Once a virtual element Health problem is identified • Single screen provides details on problem based on behavioral understanding of the element • Points to the Root Cause metrics to help you troubleshoot • Eliminates 100s of clicks and memorization of many metric behaviors that 1st generation monitoring tools require Health and Workload together tell you a lot • Workload High & Health High – Normal Behavior for this timeframe • Workload High & Health Low – Something is amiss! 6 Important Note Low Health does not imply a problem. It tells you that the object is acting differently than normal. Understanding your Virtual Environment - Capacity Capacity Measures • How much time do you have left before a object runs out of resources? • Based on a 0-100 scale – Higher the number the longer you have • Thresholds User Configurable • 30 Days Left = RED • 60 Days Left = Orange • Etc. Capacity measured for critical resources • CPU • Network I/O • Storage I/O • Memory Capacity Details View • Shows the chart and trend for each of the above resources • Denotes current state • Projected breach point and days left 7 Business Benefits 8 Increased Visibility • Lack of holistic VC environment view • Single pane of glass • Can’t determine state of all elements • All VC data contextually consolidated (clusters, hosts, guests) at once • Overwhelming details obscure valuable One click to any detail • Filters on “all” “normal” and “problem” Searches on any string. information. BEFORE AFTER • Visibility, comprehension of virtualized environment in one screen • Better product usability • Visually isolate problems via a “HUD” for vCenter • Unnecessary details hidden until necessary. 9 Reduced Complexity • Administrators blind to brewing • Provide a single measure of normality across all virtualized elements – Health problems • Too much data, too many clicks • Preset thresholds, many details • Automatically aggregate, correlate • Impossible to understand health of elements states of 100s of metrics into two scores for each element – Health and Workload BEFORE • Reduce complexity of usage • Remove guesswork, provide clarity into the environment • Speed up MTTR • Enable administrators to do more with less. 10 AFTER Slide 10 Understand Normal Metric Behavior • Unable to understand normal range of metrics • Is 65% usage normal for an hour, day, week or month? • Visibility into normal operation of every metric in VC Slide 11 • Continuous, automatic learning of normal behavior • Or, is it the beginning of a problem? BEFORE AFTER • Understand metric behavior based on history • Project forward future behavior hours or days in advance • Remove guess work and confusion, clarify expectations • Equivalent of 10 people watching, measuring and adjusting system constantly. 11 Workload Optimization • VC unaware of affinities and workload profiles of all VMs • Only understands raw resource • Calculates and stores workload profile of each ESX • Increase density by matching opposite VM behaviors on an ESX consumption • Ensure smooth, consistent use of resources BEFORE • Increase density of VMs per ESX • Optimize use of resources • Consistent and maximized ESX workloads 12 AFTER Slide 12 Understand Impact of Change • Change is common and necessary in VM environments • Change can lead to degradation in • Changes and events mashed on health chart for every element • Easier to see impact of change and before and after performance performance BEFORE AFTER • Immediate visibility into impact of change • Visual correlation to component's health • Admin can immediately determine if change had positive (expected) or negative (unexpected) effect on the element 13 Slide 13 Multidimensional Analysis • Which of my many Hosts have high levels of CPU Ready contention but low memory usage? BEFORE • Slice, dice, visualize entire environment by any of 100s of VCcollected metrics AFTER • Full Business Intelligence like capabilities • Slice and dice historical collected data across any dimension • Visualize results in heat maps, single click drill down to resource details. 14 Slide 14 Screenshots 15 Performance dashboard based on self-learning analytics Visualize environment performance in three unique dimensions Simple, actionable scores that indicate overall performance Highlights resources that are deviating from “normal” behaviour 16 Get “At-a-glance” insights into performance issues Performance scores “Details” for further analysis Visualize impact 17 Drill down into problem source Key metrics of interest based on continuous learning of “normal” behavior Stress caused by net I/O Quickly identify problem source 18 Correlate cause-and-effect of the problem Check health of related objects in the hierarchy Correlate events that occurred at the same time 19 Deep Dive into Disk and Network IO performance Disk subsystem performance details by datastores and LUNs 20 Network statistics for every NIC Identify and isolate KPI metrics Quickly identify “suspect” performance metric KPI history with timestamp to indicate root cause 21 Anticipate Capacity Issues Before They Happen Proactive warning related to capacity shortfall Correlated workload metrics forecast a potential breach 22 Project forward future issues hours or days in advance Opportunities to remediate Move VMs to another host? This host looks healthy… 23 This host seems to be overloaded! Individual performance metric details Single view that correlates multiple metrics Detailed list of all metrics indicating smart alerts 24 vCenter Operations Architecture, Process and Deployment 25 vCenter Operations Standard Architecture Four Main Services: Collector, Analytics, Web, ActiveMQ Architecture includes PostgresSQL DB File-based DB (FSDB) for raw metric storage Single Collector for vCenter embedded in appliance 26 vCenter Operations Standard Processing 1a: vCenter Collector collects metrics, topology & change events from vCenter 3: Incoming data points are tested against Dynamic Threshold bands and used 2a:toAnalytics daily to calculateruns Health, determineand hour-by-hour Workload Capacity Dynamic Thresholds for next 24 hours 4: Results provided to UI: Update “Badges”, provide Root Cause for Health scores, etc. - Ongoing - 27 1b:isData 2b: Full FSDB scanned stored in by the analytic algorithms to determineFSDB per metric best match the next 24 hour period 2c: Store metric Dynamic Thresholds data in PostgresSQL DB VMware vCenter Operations Standard - Deployment One vCenter Operations Standard per vCenter instance For VMware environments of 1500 or fewer Virtual Machines vCenter Operations Standard is a virtual appliance (.ova) • SUSE Linux Enterprise Server 11 SP1 • 8GB RAM • 2 vCPUs • 124 GB Disk (4 GB system disk + 120 GB data disk) Supported Systems • ESX host where the appliances is deployed to must be 4.0 U2 and above • 4.1 is recommended • vCenter • vCenter 4.0U2 • vCenter 4.1 – Preferred as more data is available 28 VMware vCenter Operations Standard - Deployment Simplified implementation – 15 mins • Deploy the appliance – Deploy OVF Template • Change passwords and set Timezone • Set up network configurations (Optional) • Connect to vCenter Server • IP, Admin User Name, Admin Password, Collector User Name, Collector Password • Apply your license Polling and analytics start automatically • Polling set to every 5 mins Accessing the UI • Supported browsers include: Internet Explorer 7 or 8, or Firefox 3.6.x • Internet Explorer 7 is required on the machine where vSphere Client runs 29 vCenter Operations Editions 30 VMware vCenter Operations Editions vCenter Operations Enterprise + Full Configuration & Compliance Management vCenter Operations Advanced vCenter Operations Standard + Capacity Planning + Other VMware & 3rd Party Integrations (View, management, servers, storage) Performance Real-time Capacity Configuration Change vSphere VMware Cloud / vCenter 31 Non-VMware (incl. physical) environments Understanding the vCenter Operations Editions Function Scope vCenter Operations Standard Edition 32 vCenter Operations Enterprise - Stand-Alone Data Sources vCenter x 1 • Any 3rd party monitoring tools’ time series data • Change events • Multiple vCenter Servers Objects vCenter Objects (i.e.) • Data Centers • Clusters • ESX Hosts • Datastores • VMs x 1500 Unlimited Scope (i.e.) • Applications • Network Infrastructure • Storage • Hosts (ESX, Win, Linux, etc) • VMs Users Infrastructure (e.g. VI Admins) Operations, Infrastructure, Application Teams, Business Owners, CxOs Dynamic Thresholds Yes Yes Performance Root Cause Yes Yes Proactive Alerting No Yes Customizable Dashboards No Yes Notifications No Yes Demo 33 Questions 34