WEB11 WebSphere extreme Scale et WebSphere DataPower XC10 Appliance : les solutions de caching élastique WebSphere Catherine Ezvan Consultante WebSphere IT Specialist certifiée Correspondante IBM auprès du Guide Share France WebSphere cath.ezvan@fr.ibm.com
Disclaimer Information regarding potential future products is intended to outline our general product direction and it should not be relied on in making a purchasing decision. The information mentioned regarding potential future products is not a commitment, promise, or legal obligation to deliver any material, code or functionality. Information about potential future products may not be incorporated into any contract. The development, release, and timing of any future features or functionality described for our products remains at our sole discretion.
IBM Vision at a Glance Smarter Planet instrumented interconnected intelligent Green & Beyond New Intelligence Dynamic Infrastructure Smart Work
Application Infrastructure Portfolio Adoption Patterns IBM Offerings Application Foundation WebSphere Application Server Family WebSphere smash IBM CICS Transaction Server Intelligent Management WebSphere CloudBurst Appliance WebSphere Virtual Enterprise Extreme Transaction Processing WebSphere extreme Scale WebSphere DataPower XC10 WebSphere Real Time WebSphere Compute Grid
Standard Online Transaction Processing Model Users Presentation Static (HTML, JPG, GIF) Logic Java (Servlets, EJBs) Data Relational (IBM DB2, etc.) Dynamic (Servlets, JSPs) Traditional (COBOL, C/C++) Non-Relational (CICS, IBM IMS, etc) Other (VSAM, etc) This is the three tier model you may have heard about It may be physically three tier or physically one tier Logically it s three tier that s the key Works and works well as long as the demands placed on it don t outstrip some element of this which often happens when scaled up
Internet Has Opened Up a World of Users Users Compute Resources Think about some of the very large models in today s world: Online banking Online retailing Auction sites Social networking sites Potentially multi-millions of simultaneous users These numbers put enormous strain on the resources need to scale We re using very large here to set the stage to explain why data grids came about. However, extreme Scale can provide cost saving value in any sized environment.
Scaling the Presentation and Logic Layer Here s what we often see in response to work demands overwhelming some element of the design: Logical three tier physical one or many Users Data Relational (DB2, etc.) Non-Relational (CICS, IMS, etc) Other (VSAM, etc) Presentation and Logic Scaled Out Scaling out the presentation and logic layers spreads the work The data layer is a another matter more difficult to spread out
Scaling the Data Layer Logical three tier physical one or many Users See legend below Share the data -- this is what Parallel Sysplex does Works well proven technology. One physical copy shared between data resource engines with Sysplex management of the data locks Mirror the data -- various data mirroring technologies do this Multiple copies of the data updates to one reflect over to the other Partition the data -- split it across databases Split the data in some fashion and have application go to appropriate data partition
Caching Less-Volatile Data When scaling limitations are felt, one of the first things pursued is the caching of data that doesn t change much. General principle: avoid data call if possible. Users Cached Data Less I/O activity against the data subsystems Cached Data Example: basic user profile data for online banking customers or profile information for social networking site. Doesn t change much but it does change sometimes So the invalidation/update model has to be taken into account
Application Topologies Today Web Server Tier Application Server Tier Database Tier
Modern Application Infrastructure Topology with WebSphere Web Server Tier App Server Tier Elastic Data Grid Database Tier 1 extreme Scale for maximum flexibility covering data and application oriented scenarios 2 XC10 IBM HTTP Server WebSphere Application Server DataPower XC10 for simple data oriented scenarios: HTTP Session Replication Elastic Dynacache Web Side Cache DB2 UDB
Innovative Elastic Caching Solutions Data Oriented Session management Elastic DynaCache Web side cache extreme Scale Ultimate flexibility across a broad range of caching scenarios In-memory capabilities for application oriented scenarios Petabyte analytics Data buffer Event Processing Worldwide cache In-memory OLTP In-memory SOA Application Oriented DataPower XC10 Appliance Drop-in cache solution optimized and hardened for data oriented scenarios High density, low footprint improves datacenter efficiency Elastic caching for linear scalability High availability data replication Simplified management, monitoring and administration
What is WebSphere extreme Scale? A distributed caching platform. It takes the free memory across a number of Java Virtual Machines (JVM ) and marshals them into one logical cache. New York San Francisco London Shanghai Scales from simple inprocess topologies to powerful distributed topologies. Can be used as a form of in memory database to manage application state. It can be used as a platform for building powerful Extreme Transaction Processing (XTP) / Data Grid applications.
Traditional Cache Operation Cache capacity determined by individual JVM Size. Size of each cache = M # JVMs = N Total cache = M Invalidation load per server increases as cluster grows. Cold start servers hit the database. Application Application Application Application A A A New Server with cold cache Invalidation Chatter Invalidation load increases with cluster size Database Application A High load on Database Redundant copies of data at different versions
WebSphere extreme Scale Cache Operation Cache capacity determined by total cluster size Size of each cache = M # JVMs = N Total Cache = M x N Application Application E A B E No invalidation chatter Linearly scalable Less load on database and no cold start spikes Application B A Application C D Cache 5x Larger Cache 4x Larger Application D C
WebSphere extreme Scale and Distributed Caching: A Primer 3 2 Data is replicated automatically so high availability is automatic DB Data is kept in sync with database automatically (using adapters and technologies like JPA and Hibernate) 1 JVM JVM JVM JVM Data is stored in memory across a set of distributed machines. These machines can be across data centers or geographies. Distributed Cache Application 5 4 As data grows, just add more JVM s Applications use APIs to access, modify and update data -or- they can use a plug-in adapter (e.g. Dynacache for WebSphere Application Server)
Tell Me What It Really Means To Applications! DB JVM JVM JVM JVM JVM Application Application Application Consistent application and transaction response times even as data grows! This is referred to as linearly scalable No need to continually re-architect as data and transaction volumes continue to grow and grow Much faster access to data than a traditional database (just a hop over the network) Most of the management of data is automatically handled by W ebsphere extreme Scale WebSphere extreme Scale
Interesting! Tell me More!! DB Apps & Data Application JVM Application JVM Application JVM Application JVM Application JVM Imagine what happens when the application lives with the data? The data is hosted in a JVM. Which could be WebSphere Application Server or ANY Java EE application server. Or it can just be a JVM (no application server) any Java SE 1.4+ compliant JVM. The application lives there too. This is called co-location. The application and data are together. It s a marriage! It results in very good application and transaction performance because the application lives with the data, which is in-memory. New application patterns are possible. Cannot get better application / transaction performance than this WebSphere extreme Scale
Simple Caching Scenarios Challenges Application makes redundant calls, doing something over and over again, on expensive back-end systems Generally, to access data that does not change much (e.g., user profiles) Offload Redundant Processing Benefits Free up expensive back-end systems for critical tasks Reduce costs of system cycles for repetitive data retrieval Increase performance through inmemory, network cache Challenges Web sites that need better management and automatic fail-over of Web sessions usually WebSphere Commerce, WebSphere Portal or retail-related sites Drop-In HTTP Session Replication Benefits Automatic drop-in WebSphere extreme Scale cache without invasive coding changes Higher availability and performance for revenue-producing applications Challenges Web applications that use (WebSphere Application Server) DynaCache and need better performance and scalability of their caching investment Drop-In extension for DynaCache Benefits Better performance: turbo-charge WebSphere Application Server caching layer via WebSphere extreme Scale drop-in cache with no coding changes
WebSphere extreme Scale Non-invasive middleware Lightweight Single JAR, 15MB Integrates with W ebsphere Application Server and stack products WebSphere Business Events WebSphere Commerce IBM ILOG JRules And more! Also supports Non-W ebsphere servers Geronimo, JBoss, and others Enhances existing application infrastructure with little to no application code change
WebSphere extreme Scale: It s Not Just for Java Rest Data Service Allows HTTP clients, including.net and AJAX clients, to use and exploit W ebsphere extreme Scale. Rest Service is hosted on application server such as WebSphere Application Server or Tomcat. HTTP Clients WebSphere extreme Scale.Net/WCF AJAX OData Rest Service Grid Database PHP Other
WebSphere extreme Scale: Replication and Availability Granular control over where information is placed and replicated New York Application A E London Application Application E F I J Route client transactions to information based on your specifications Application Application B C F G Application Application G H K L Flexible definition of zone Across data centers Specific chassis Rooms in a building Application D H Beijing Application I B Application Application J K A D Application L C 2008 IBM Corporation
WebSphere extreme Scale: Native Integration with WebSphere Application Server WebSphere extreme Scale works with all WebSphere Application Server 6.x and 7.x versions WebSphere extreme Scale extends the value of WebSphere Application Server deployments by providing: Session management plug-in for multi data center support Dynamic cache plug-in to turbo-charge existing caching environments JPA / Hibernate side-cache to accelerate existing database queries Session Management Plug-in Dynamic cache service Plug-in Session Initiation Protocol Support JPA / Hibernate Side-cache Global Transaction Integrity Deep Management Integration Programming Models Serviceability WebSphere extreme Scale WebSphere Application Server
WebSphere extreme Scale: Tell Me What this Really Means to my Business! 1. Benefit from extreme cost savings for your IT processing Experience rapid time-to-value (payback in as few as 3 months) Achieve high return on investment (ROI) 2. Benefit from investments in your existing assets Extend application server investments (WebSphere Application Server, 3rd party Java application servers, Microsoft.NET) 3. Benefit from better use of your scarce and expensive IT resources Offload redundant data access from databases, LDAP servers, mainframes -- freeing up valuable backend resources and capacity 4. Benefit from high performance architectures to support your growth Provide seamless cache infrastructure across applications Deliver high performance and consistent response times Ensure high availability of critical applications
Introducing IBM WebSphere DataPower XC10 Appliance Easy drop in use for common scenarios Scale out with ease Large, elastic cache allows you to scale more economically while providing high Quality of Service Fault tolerance Support for data-oriented caching scenarios without rip & replace Lower risk of data loss while providing continuous availability Flexible and simple user management Simple solution for real world management and monitoring
WebSphere DataPower XC10: Easy drop-in use for common scenarios HTTP Session Management WAS Dynamic Cache Service ( Dynacache ) support Web Side Cache Little or no code changes required!
WebSphere DataPower XC10: Get Started page links for data caches
WebSphere DataPower XC10: Offloaded session management for HTTP requests WebSphere Application Server connects seamlessly to the DataPower XC10 appliance Client code must be installed on W ebsphere Application Server systems Easily configure WebSphere applications to store HTTP session data to a data cache on the DataPower XC10 appliance through the WebSphere Application Sever administrative console Replaces other session replication mechanisms
WebSphere DataPower XC10: End-to-end Session Management Scenario WebSphere Application Server side DataPower XC10 appliance side WebSphere Application Server Integrated console 2 Install DataPower XC10 client code DataPower XC10 Appliance console 1 Install DataPower XC10 Appliance DataPow er XC10 session data 3 Configure to use data cache on the DataPower XC10 Appliance for session management 4 Session data can be monitored on the appliance web console Application DataPow er XC10 servlet agent
WebSphere DataPower XC10: Dynamic Cache service support DataPower XC10 provides client code and a plug-in for WebSphere Application Server applications to support DynaCache API Allows applications deployed to WebSphere servers to use DataPower XC10 as a drop-in cache, instead of storing cache data in local memory or multiple instances of a disk cache
WebSphere DataPower XC10: Web Side Cache Used to store data for fast, lower-cost access than a database Uses ObjectMap APIs from W ebsphere extreme Scale Every time data is needed, the web side cache on the DataPower XC10 Appliance is checked first If the value is not found (cache miss), then the data is retrieved from the backend database and inserted into the cache Client can run in a standard Java EE compliant server environment or in any Java Virtual Machine compliant with Java SE V1.4 or beyond
Form Collectives of XC10 Appliances Collectives are a set of appliances which know about each other Allows for failover recovery if one of the appliances in the collective fails Default collective contains the appliance where you logged on Add additional appliances using the console interface Add additional appliance by supplying: Host name or IP address of the appliance to be added Secret key of the appliance to be added Adding one appliance to collective that contains only 1 appliance: Provides failover capability because 2 nd appliance automatically gets replica of all data
Flexible and Simple Management Native interfaces for administration and monitoring Relief to simple real world problems! Real-time monitoring that dynamically updates with the number of entries in the grid
Built-in console for management and administration How to set up the appliance scenarios for caching
WebSphere DataPower XC10: Summary Accelerated Time to Value Reduces the time necessary for install, setup and configuration through out-ofthe-box, drop-in use for simple side cache scenarios, HTTP Session replication and WebSphere Application Server dynamic cache service Simplified management and administration Offers a built-in, simplified administration and monitoring console to enable efficient setup, configuration, and management of the appliance and transaction load within your datacenter Ensures high availability of data for mission-critical applications Scales with simplicity Delivers high performance and consistent response times
WebSphere extreme Scale and DataPower XC10: Learn More Download WebSphere extreme Scale for free and build a trial app http://www.ibm.com/developerworks/downloads/ws/wsdg/ learn.html Browse the WebSphere extreme Scale Data Sheet ftp://ftp.software.ibm.com/common/ssi/pm/sp/n/wsd14072usen/ WSD14072USEN_HR.PDF Read Rob High s Article on XTP and WebSphere extreme Scale http://www-01.ibm.com/software/solutions/soa/newsletter/jul09/ article_xtp.html?open&cm_mmc=5752-_-n-_-vrm_newsletter- _-10481_125043&cmibm_em=dm:0:16632298 XC10 Web Site http://www-01.ibm.com/software/webservers/appserv/xc10/ Getting Started Wiki - extreme Scale and XC10 ftp://ftp.software.ibm.com/common/ssi/pm/sp/n/wsd14072usen/ WSD14072USEN_HR.PDF Additional resources Weekly video podcasts covering customers questions and forum posts on the IBM WebSphere extreme Scale product. http://www.youtube.com/user/ ibmextremescale#p/a WebSphere Extreme Transaction Processing for Developers Space will discuss various topics for developing and deploying XTP applications and will point out emerging trends, benefits, challenges, and features associated with it. http://www.ibm.com/developerworks/ spaces/xtp
Backup
Offload Redundant Processing: Co-located application Business logic runs in parallel with where the data is stored. This is called co-location. The application and data are together. Results in very good application and transaction performance because the application lives with the data, which is in-memory. DB Application JVM Application JVM Application JVM Application JVM Application JVM Data Grid / Cache
Drop-In HTTP Session Replication HTTP sessions can be replicated across servers using extreme Scale A servlet filter that enables session replication can be inserted into any Web application Provides a session persistence approach that is independent of the W ebsphere cell infrastructure WebSphere products can use WebSphere extreme Scale as an upgraded session persistence mechanism Non-WebSphere servers (such as Geronimo or JBoss) can also use this servlet filter
Drop-In Dynamic cache service support Allows applications using the WebSphere dynamic cache service to leverage the advanced features and performance improvements of extreme Scale Supports W ebsphere Application Server V6.1 or V7.0 Dynamic cache evictors, dependency-based invalidation functions, and event listeners can be used on the WebSphere extreme Scale cache Dynamic cache can keep statistics for each grid instance
WebSphere extreme Scale: Monitoring Tools Support for leading external monitoring products IBM Tivoli Monitoring CA Wily Introscope Hyperic HQ Built-in monitoring with WebSphere Integrated Solutions Console and xsadmin command-line utility
WebSphere DataPower XC10: Monitoring Monitor settings: Allow users to monitor load for an application connected to the DataPower XC10 appliance Three view options available: Data grid overview Individual data grid overview Data grid detail reports