IBM Software Group Web Sphere Virtual Enterprise 2

IBM Software Group What is Web. Sphere XD? Software to virtualize, control, and turbo-charge your application infrastructure Infrastructure Optimization Intelligent Workload Management Virtualization Automatic Sense & Respond Management Data Fabrics & Caching Innovative Application Patterns (beyond OLTP) 2 Web. Sphere © 2008 IBM Corporation

IBM Software Group Web. Sphere XD Packaging Structure Available as a single, integrated package, or as 3 individual components Operations Optimization Virtual Enterprise Data Grid e. Xtreme Scale Compute Grid 3 Web. Sphere © 2008 IBM Corporation

IBM Software Group VE Virtual Enterprise - Customer Value/Scenarios Save Money through Server Consolidation Ø Consolidate multiple under-utilized servers into a virtualized pool Ø This is where the VE Value Assessment is useful Improve Environment Manageability Ø Enable the Middleware Operations team to run the environment easier Ø Application versioning and monitoring Ø Centralized Installation Management Ø This is not about saving hardware/software cost Improve Availability, Scalability and Resiliency Ø Adaptability of the VE environment Ø Traffic management Ø Better hardware utilization Ø Service policy 5 Web. Sphere © 2008 IBM Corporation

IBM Software Group Dynamic Operations Operating Environment Customer use case: Large Financial Company Challenges Ø Underutilized servers Ø Inability to share resources across server pools – especially during peaks Ø Inconsistent quality of service for business critical applications 50% 0% 15% Utilized Servers Portfolio Forecasting 10% Utilized Servers 0% 50% Ø Human Intensive Monitoring and Managing Environment Account Management 100% Ø 100 application servers 50% Ø 30+ applications 75% 55% 100% 20% Utilized Servers 0% Environment 100% Conventional Distributed Environment 100% Stock Trading 6 Web. Sphere © 2008 IBM Corporation

IBM Software Group Dynamic Operations Operating Environment Customer use case: Large Financial Company Stock Trading Conventional Distributed Environment Customer Support Virtualized Ø Pool Resources (Node Groups) Ø Virtualized Applications (JAS) Account Management Autonomic Ø Operational Policies are attached to Application to reflect operational goals and importance of application Ø Autonomic Managers monitor environment for maximum utilization using business goals Risk Management 0% Ø Reduce total cost of ownership (doing more with same/less) ØIncrease stability and repeatability of Environment 7 Web. Sphere Utilized Servers 50% Results 100% Portfolio Forecasting 55% RESOURCE POOL © 2008 IBM Corporation

IBM Software Group Web. Sphere Virtual Enterprise Standard Topology 8 Web. Sphere © 2008

IBM Software Group Defining SLAs through Service Policy is definition of a performance goal used by VE to decide how to manage resources in the server environment Defined in terms of the end user result the customer wishes to achieve Comprised of three parts: A set of classification rules to decide which policy applies to a given request A performance goal the user desires to be achieved This is a response time based goal (i. e. 500 ms average response time for all requests) An importance level to inform VE of the relative priority of different classes of work 9 VE does not cause starvation but will handle different workloads with different priorities Web. Sphere © 2008 IBM Corporation 9

IBM Software Group Techniques to Meet SLAs VE two primary techniques to meet Service Policy objectives Ø Traffic Shaping and Application Placement Traffic Shaping Ø Not all requests are equal Ø Serving work first-come-first-serve is not necessarily the best approach Ø Controls Traffic in a number of ways • Prioritization – Processed in order of importance • Flow Control – Using queuing, the rate of work being sent to the server cluster is controlled • Traffic Spraying – Dynamic Workload Management • Overload Protection – Control total amount of outstanding work for each class of service Application Placement Ø The ability to adjust the size of a Dynamic Cluster in real-time Ø Controls how much capacity is online for an application at any moment in time Ø Integration with Tivoli Intelligent Orchestrator to provision additional resources into VE 10 Web. Sphere © 2008 IBM Corporation

IBM Software Group Dynamic Operations Overview Dynamic Operations enables set up, monitoring and management for multiple applications sharing server resources. A service policy is used to specify response time goals application requests must meet. System resources are managed based on current workload, and relative importance of requests. The On-Demand Router is used to route work to application server nodes Ø Enhanced version of the Proxy Server Ø Controls request prioritization, flow, and routing in an Virtual Enterprise (VE) environment 11 Web. Sphere © 2008 IBM Corporation

IBM Software Group Virtual Enterprise Architecture Overview Classification Prioritization and Routing and Flow Control Load Balancing (1 per ODR) (1 per Dynamic Cluster) Application Placement Controler (1 per cell) Medium Importance Low Importance Financial Advice FA AM Node FA 5 ODR AM Node FA 4 Account Mngmt AM Node ST 3 Stock Trading Node ST 1 Node ST 2 High Importance AM Provisioning Executions Stock Trading Account Mngmt Financial Advice Application Demand Resource State Operational Policy 12 Web. Sphere VE Decision Makers Provisioning Decisions © 2008 IBM Corporation

IBM Software Group Application Editions Edition Management of an application on Web. Sphere nodes Ø Interruption-free rollout of application updates Ø Ability to “roll back” to a previous application version Ø “Validation mode” to verify functionality using a subset of users Traffic management and routing on non-Web. Sphere nodes Ø Representations of existing applications must be created in VE Ø Can be edition-qualified such that routing policies can be configured Ø Specify how traffic should be managed and routed to multiple editions 13 Web. Sphere © 2008 IBM Corporation

IBM Software Group Operational Scenarios: Grouped Rollout On-demand routers Edition 1. 0 2. 0 quiesce & stop restart application Edition 1. 0 requests Edition 1. 0 Dynamic cluster 14 Web. Sphere © 2008 IBM Corporation

IBM Software Group Operational Scenarios: Atomic Rollout On-demand routers Edition 1. 0 2. 0 quiesce & stop Edition 1. 0 2. 0 application restart requests request Edition 1. 0 2. 0 quiesce & stop Edition 1. 0 2. 0 request Dynamic cluster 15 Web. Sphere restart © 2008 IBM Corporation

IBM Software Group Validation Mode Creates a ‘clone’ of the deployment target (Dynamic Cluster) in production environment and deploys new version Use routing policies to control edition visibility Ø Restrict new edition to only certain users, for example Use rollout function to move edition from validation mode to production Ø Edition deactivated on ‘clone’ environment Ø Edition rolled out on original deployment targets 16 Web. Sphere © 2008 IBM Corporation

IBM Software Group Operational Scenarios: Validation Mode Routing rules Edition 1. 0 Dynamic cluster ‘DC 1’ On-demand routers clone Edition 2. 0 Legend: Edition 1. 0 Requests Dynamic cluster ‘DC 1 -Validation’ Edition 2. 0 Requests 17 Web. Sphere © 2008 IBM Corporation

IBM Software Group Operational Scenarios: Concurrent Activation Routing rules Edition 1. 0 Dynamic cluster On-demand routers Edition 2. 0 Legend: Edition 1. 0 Requests Edition 2. 0 Requests 18 Web. Sphere Static cluster © 2008 IBM Corporation

IBM Software Group Application Management • • 19 VE can begin to route and manage traffic for other middleware servers when representations of existing applications are created into VE These application representations can be edition-qualified such that routing policies can be configured to specify how traffic should be managed and routed to multiple editions Web. Sphere © 2008 IBM Corporation

IBM Software Group On Demand Router – Traffic management and routing The On Demand Router (ODR) is a component that logically replaces and extends the functionality of the ND HTTP Plug-in The ODR provides the standard functionality of a proxy server with added On Demand features Ø Request classification and prioritization Ø Request queuing Ø Routing and load balancing Ø Weighted round robin dispatching with Dynamic WLM weights Ø Dynamic routing table updates with multiple Web. Sphere backend cells Ø HTTP Session affinity Ø SSL ID Affinity Ø WPF Partition Affinity 20 Web. Sphere © 2008 IBM Corporation

IBM Software Group Monitoring Operations Monitoring a dynamic goals directed environment Operations tabs Ø Request management Ø Co. Location (Stability, Weight, CPU) Ø Active tasks Reports tab Ø Select metrics Ø Chart groups 21 Web. Sphere © 2008 IBM Corporation

IBM Software Group Health Management – Health Policies Health policies can be defined for common server health conditions Health conditions are monitored and corrective actions taken automatically Notify administrator Ø Capture diagnostics Ø Restart server Ø Application server restarts are done in a way that prevent outages and service policy violations 22 Web. Sphere • Health Conditions • Age-based: amount of time server has been running • Excessive requests: % of timed out requests • Excessive response time: average response time • Excessive memory: % of maximum JVM heap size • Memory leak: JVM heap size after garbage collection • Storm drain: significant drop in response time • Workload: total number of requests © 2008 IBM Corporation

IBM Software Group Custom Health Conditions VE enables customers to create expressions defining what “unhealthy” means in their environment Custom expressions can be built using operands which represent metrics from the On Demand Router, base PMI metrics (WAS only), MBean operations and attributes (WAS only), and/or URI return codes. Complex expressions using a mix of operands is supported. Other middleware server types can leverage the ODR metrics and URI return code operands Create the health policy by using the create. Health. Policy Admin. Task command 23 Web. Sphere © 2008 IBM Corporation

IBM Software Group Custom Health Actions Customer often have needs for specific actions to occur when an unhealthy situation is detected. VE provides flexibility to customers by allowing them to define custom actions in addition to a standard set VE provides and allow them to define an action plan to be carried out when the unhealthy situation is detected. 24 Web. Sphere © 2008 IBM Corporation

IBM Software Group What are Tasks? Tasks are generated by VE runtime components to alert the operator to take action or be aware of certain situations in the runtime environment Tasks are generated by the Health Management Controller and Application Placement Controller when they are in supervised mode Planned Tasks Ø Have a recommended course of action for the user to take (Action Plan) Non-Planned Tasks Ø Alert the operator to the situation, but have no recommended action plan 25 Web. Sphere © 2008 IBM Corporation

IBM Software Group Tasks What makes up a task? Ø A global severity depicting how serious the task is Ø The submitter id declaring what runtime component sent the task Ø A global task id for unique identification Ø A task explanation describing why the task was submitted to the user Ø Lifecycle – very basic for Non. Planned to simple for Planned Ø Task targets and monitors • What the task is specifically targeted towards • What the user should look into in order to verify the task Task management interaction Ø Automated reactions presented as runtime tasks Ø Task severity increases over time Email Notifications can be sent out upon new task arrival Ø By default, disabled Ø New Tasks only • Renewed, Suppressed, Expired, etc… do not trigger an email 26 Web. Sphere © 2008 IBM Corporation

IBM Software Group Server Maintenance Mode VE provides the capability to isolate a running server (of any type) from production traffic. This allows for problem determination to be performed on the server or other maintenance without disruption to production traffic. If the server is a member of a dynamic cluster, a new cluster member will first be started before the server is placed into maintenance mode in order to assure the minimum policy on the dynamic cluster is met. 27 Web. Sphere © 2008 IBM Corporation

IBM Software Group Highly Available Deployment Manager Configuration Each deployment manager on a separate machine Ø Only one is active Ø Others are standby Shared file system required for dmgrs to share configuration repository Ø File system with recoverable locks required - e. g. SAN FS, DFS v 4 JMX traffic proxied through VE On-demand Router (ODR) Ø SOAP connector only HA ODRs recommended Ø (they’re recommended for production VE configurations anyway) hadmgr. Config command line utility provided to perform configuration 28 Web. Sphere © 2008 IBM Corporation

IBM Software Group Highly Available Deployment Manager “Warm-standby Model” 29 Web. Sphere © 2008

IBM Software Group Highly Available Deployment Manager Take-over after primary failure… 30 Web. Sphere

IBM Software Group Repository Checkpoint/Restore Offers ability to restore checkpoints on ‘live’ Deployment Manager, unlike existing restore. Config utility Offers ability to administratively save copy of full configuration repository Ø Improved preparedness for disaster recovery Offers ability for administrative system to automatically create ‘delta’ checkpoints Ø checkpoint consisting of backup copies only of files changed by a discrete configuration update Ø Undo-capability to unwind the last ‘n’ configuration updates Ø configurable stack depth 31 Web. Sphere © 2008 IBM Corporation

IBM Software Group Repository Checkpoint/Restore 32 Web. Sphere © 2008 IBM Corporation

IBM Software Group Middleware Servers and Applications XD 6. 0. x already supports non-Web. Sphere environments (MSE) Ø Mostly for Advanced Routing/Traffic Control and Monitoring VE 6. 1 expands the non-Web. Sphere Support to include most critical VE functions Ø Application Placement through Dynamic Clusters Ø Health Management Ø Application Edition Management VE 6. 1 unifies the management model into a new system called Middleware Servers and Applications 33 Web. Sphere © 2008 IBM Corporation

IBM Software Group Unified Administration across Middleware Common collection views allow administrators to interact with their Web. Sphere servers side by side with servers from other middleware platforms 34 Web. Sphere © 2008 IBM Corporation

IBM Software Group First Class Support for Non-Web. Sphere Platforms Three categories of support for middleware server types… Complete Lifecycle Management • • • Create/remove server instances Govern all aspects of server configuration Provide operational control Deploy applications Server health and performance is monitored and visualized. . 0 , 7 6. 1 , 2. 0 V V 2 Application Server Network Deployment Assisted Lifecycle Management • • • Provides specific templates for creating representations of existing servers and applications Servers can be controlled operationally Server health and performance is monitored and visualized. , 5. 1 Application Server Community Edition 6. 0 V Application Server Base & Network Deployment Tomcat . 1 V 1 Application Server Community Edition Generic Lifecycle Management • • Provides generic templates for the user to manually define servers and operational commands. Control server operations and monitor health and performance … 35 Web. Sphere © 2008 IBM Corporation

IBM Software Group Web. Sphere Virtual Enterprise étend la platforme Web. Sphere Applications ISV Les produits de la plateforme Web. Sphere i. e, Web. Sphere Process Server, Web. Sphere Portal, etc. Web. Sphere Application Server Network Deployment (WAS ND) WAS for z/OS Etend la Qualité de Service pour la plateforme Web. Sphere Virtual Enterpise • Packagé en tant que complément à l’infrastructure Web. Sphere Application (Installation d’un simple delta à l’environnement existant) • Complètement intégré à l’environnement d’administration de Web. Sphere 36 Web. Sphere © 2008 IBM Corporation