Transcription

Platform Rationalizationwith Private Cloud and Oracle Engineered Systems

Authors: Johan Louwers and Sanjiv NashtePrefaceToday’s organizations use only a fraction of the technologythat makes up their infrastructure, leading to high costsand inefficiency. The organic growth of many enterprise ITfootprints has resulted in costly and hard-to-support-andmaintain environments.Due to the nature of these organically grown IT footprints,oftentimes the majority of IT budget is spent on runningand maintaining the existing landscape instead of beingused for improvement and innovation.With its platform rationalization program offering,Capgemini can help you rationalize your entire IT footprintand, as a result, lower your total cost of ownership. Bydecreasing the total cost of ownership, budget can befreed for future optimization and innovation to help improveyour day-to-day business processes.2Platform Rationalization

Oracle Based Big Data Strategy the way we see itTable of Contents12Platform Rationalization41.1 Non-rationalized IT landscapes41.2 Reasons for rationalization41.3 Rationalization goals4Capgemini Approach52.1 Capgemini project blueprints52.2 Capgemini deployment bluprints52.2.1 Private cloud-based deployments52.2.2 Application consolidation62.2.3 Database consolidation62.2.4 Datacenter consolidation63Cloud Deployment Architecture83.1 Cloud concepts83.1.1 Private cloud83.1.2 Public cloud83.1.3 Hybrid cloud3.2 Private cloud models893.2.1 Single-tenant private cloud93.2.2 Multi-tenant private cloud93.2.3 Oracle engineered systems93.3 Cloud as a Service models93.4 Infrastructure as a Service103.5 Platform as a Service103.5.1 Middleware as a Service113.5.2 Database as a Service124Conclusion143

1 Platform RationalizationWith platform rationalization, enterprises take a snapshotof the current state of their IT landscapes, mapping to realbusiness objectives, and conduct project-based rationalizationof it. Rationalization includes consolidation, optimization, andstandardization to ensure future-ready, stable, and costeffective service to the business.1.1 Non-rationalized IT landscapesIn large enterprise IT landscapes, you typically find hundredsof applications with overlapping business processes andfunctions. It is not uncommon for multiple systems to provideoverlapping services to the business even though they arebased upon a different technology stack. Operating multiplesystems with the same purpose and at the same time itresults in unnecessary costs related to staff employment andeducation.Enterprise IT landscapes usually grow in an organic manner,which unintentionally results in a diverse collection oftechnological solutions and implementations. Examplesinclude: Multiple applications providing similar business processes and functionsDeployment of large and diverse middleware portfolios,such as: IBM Websphere, JBoss, and Oracle WeblogicDeployment of large and diverse database portfolios, suchas: Oracle database, Microsoft SQL, Oracle MySQL andIBM DB2Deployment of large and diverse operating systemportfolios, such as: Redhat Linux, Oracle Linux, MicrosoftWindows and IBM AIXA wide variety of hardware implementations for compute,storage, backup and network facilitiesBecause of this large variety of software and hardwaresolutions, maintenance costs often take up a large portionof the IT budget, leaving only a small amount of budget forinnovation and improvements.4Platform Rationalization1.2 Reasons for rationalizationTypical incentives for enterprises to rationalize their ITlandscapes and, in particular, their application landscapes,include: Reduced maintenance costsDecreased complexityImproved business processes and functionsImproved stability and availabilityMore room for innovation and improvements1.3 Rationalization goalsIn a rationalization project, the primary goals are to ensurethat an enterprise IT organization is future-ready, to establishstable and cost-effective services to the business, and toenable continuous innovation to support company growth.With rationalization, this can happen thanks to:: Reduction of redundant applications, business processes,and functions Modernization of remaining applications, business processes, and functionsConsolidation of technologies into a service-based conceptStandardization of technology platformsEstablishment of a future- ready and agile IT landscapeDecreased costs and improved services that allow forgreater innovation

Oracle Based Big Data Strategy the way we see it2 Capgemini ApproachCapgemini adopts a standardized approach to platformrationalization. Capgemini uses the most appropriatemethodology for each particular situation, leveraging itssignificant global experience and expertise in businessprocess, application, database and platform rationalization.All approaches have their foundation in the Capgemini WideAngle Application Rationalization Program,or WARP, methodology.2.1 Capgemini project blueprintsFor platform rationalization, Capgemini uses project blueprintsthat are standardized and industrialized to ensure rapidresults and a rapid return on investment. The CapgeminiWARP methodology is the foundation of Capgemini’soverall approach and project blueprints. WARP is a triedand-tested approach that takes enterprises on a journeytowards a rationalized and modernized IT application portfolioin as little as seven weeks. Simply put, WARP enablescompanies to take the first step in regaining control of theirapplication landscapes.library and are updated based upon newly emerged bestpractices, technological changes, and recent productsand solutions.2.2.1 Private cloud-based deploymentsFor Oracle-centric enterprise landscapes, Capgemini hasa number of technical, functional, and project blueprints forplatform rationalization. All platform rationalization blueprintsfor medium and large enterprises revolve around the use ofprivate or hybrid cloud strategies that are primarily focused onDatabase as a Service (DBaaS) and Middleware as a Service(MWaaS) solutions within cloud computing.By implementing private cloud solutions with a foundationin Oracle engineered systems for both database andmiddleware, companies have the option to build therationalized platform and application landscape on astandardized, available, high-performance platform while atthe same time providing flexibility and agility to the businessand to developers.WARP supplies concrete facts and insight that serve as thetools for making informed decisions about how to simplify,industrialize, standardize and renew application landscapesand IT processes.Because enterprise IT landscapes often grow in an organicand evolutionary fashion, it is not uncommon for an enterpriseIT landscape to resemble the representation in figure 2.1. Thistype of landscape is diverse and costly to maintain, requiring alarge, skilled staff.WARP comprises four “streams” supported by three“engines.” These streams are PATH (vision, architecture andsolution), BIZZ (business analysis), CASE (the business case– from both a financial and value perspective) and PLAN(change planning, change scenarios and the roadmap).Experience shows that the effectiveness of any rationalizationprogram is improved considerably by the presence of thesefour streams.By rationalizing platforms and applications and implementinga standardized private or hybrid cloud-based approach as isrepresented in figure 2.2, enterprises can significantly reducecomplexity. Thanks to standardization , the IT landscapebecomes much easier to maintain thanks to a lower numberof staff required. At the same time, standardization increasesthe agility of the enterprise thanks to the flexible nature ofcloud-based solutions.The three engines – APPS, INFRA and AMBI – are highlyindustrialized ‘lenses’ that are employed to analyze theapplication landscape, the infrastructure components, andalso ambient factors. While the APPS engine provides anindustrialized analysis of the application portfolio, INFRAidentifies rationalization opportunities in the infrastructurelandscape. AMBI, on the other hand, goes beyond technologyand infrastructure and analyzes the context in which theapplication is located, including the relationship between thebusiness, the processes, and data harmonization.In figure 2.2 the foundations of both databases andapplications are based upon a private “as a Service” cloudmodel and have a technical foundation in Oracle engineeredsystems. Oracle engineered systems provide extremeperformance and form the basis for an Oracle-centric privatecloud strategy.2.2 Capgemini deployment bluprintsCapgemini maintains a large set of deployment blueprints thatform the basis of an enterprise deployment. The deploymentblueprints are part of Capgemini’s enterprise architectureOracle Enterprise Manager is the standard monitoring andmaintenance tooling from Oracle that integrates with mostOracle software and hardware and a growing number ofnon-Oracle products. In the above model, Oracle EnterpriseManager is used as the only management and monitoringtooling, which provides a single point of control for hardwareand software. At the same time, Oracle Enterprise Managerprovides self-service portals for users to ensure a full cloudbased self-service concept that can be adopted within the5

Figure 2.1: Traditional evolutionarily grown IT footprintManagement &Monitoring toolingManagement &Monitoring toolingManagement &Monitoring toolingManagement &Monitoring toolingManagement &Monitoring toolingManagement &Monitoring toolingApplicationApplicationApplicationMS SQL ServerOracle 11GMySQL ServerJBOSSOC4JTomcatWindows ServerOracle LinuxDebianRedhat LinuxSun SolarsWindows ServerDellHPDellHPSunDellDatabase LandscapeDatabase LandscapeFigure 2.2: Modern architected footprintOracle Enterprise Manager 2COracle OptimizedApplicationMS SQL Server6Oracle 11GMySQL ServerOracle OptimizedApplicationOracle WeblogicOracle Linux / SolarisOracle LinuxOracle ExadataOracle ExalogicDatabase LandscapePlatform RationalizationDatabase LandscapeOracle OptimizedApplication

Oracle Based Big Data Strategy the way we see itenterprise and includes metering and chargeback solutions forpay-per-use model billing to different business units.2.2.2 Application consolidationApplication consolidation concerns the consolidation ofmultiple standalone applications into a single operatingsystem holding multiple applications. Note that mergingbusiness functionalities from multiple applications into a singleapplication is considered to be application modernizationrather than application consolidation.A Capgemini best practice for application consolidation ofOracle-centric applications concerns consolidation ontoan Oracle Exalogic or Sparc Supercluster based upon theapplication type. For applications that cannot be deployedon these Oracle engineered systems, the Capgemini bestpractices advises consolidation on an Oracle Private CloudAppliance.2.2.3 Database consolidationDatabase consolidation concerns the consolidation ofmultiple standalone databases into a single operating systemholding multiple databases. Database consolidation is oftenconsidered when schema consolidation is not an optiondue to the nature of the content of a database and when thedatabase load allows a shared operating system. Capgeminihas a specific standardized approach around databaseconsolidation that can be executed in a standalone manner orcan be part of an overall platform rationalization program.When consolidating databases onto the Oracle Exadataplatform, it is important to understand the complexity ofdeciding where to deploy a database inside the OracleExadata machine and determining which databases is neededto implement Oracle Real Application Cluster (RAC) clusteringwhen this has not already taken place.A full rack of Oracle Exadata provides eight compute nodes,all of which can host multiple databases. Some of thequestions that need to be asked when deciding where todeploy a database include: Do the combined average system utilization and the combined average top utilization stay below the maximumthresholds of a node?Are business-critical databases spread evenly over thedifferent nodes to prevent an “all critical systems down”situation in case of a node failure?Are business-critical databases using RAC configuration inorder to allow rolling upgrades and prevent unavailability ofdatabases during patching cycles?Are the different RAC nodes of a RAC cluster placed ondifferent Oracle Exadata nodes?Are databases that need to be on different physical nodesfrom a legal and compliancy point of view placed on thecorrect nodes? Is Oracle Resource Manager configured correctly to provideensured resource capacity to business-critical databasesand processes based upon business criticality at specificmoments in time?The list of questions and considerations is a short exampleof a much wider list of questions and considerations that areincluded in the Capgemini methodology to ensure the correctdistribution of databases across nodes within an OracleExadata machine.When a consolidation project does not take into account theseand other questions during the execution of the consolidationproject, results will be sub-optimal and project goals will notbe met.2.2.4Datacenter consolidationIn the last decade, consolidation of IT landscapes has been akey performance indicator for CIOs and CTOs.In parallel, there has been a significant increase in computingpower, now capable of accommodating the most demandingworkloads enterprises could possibly run. At the sametime, enterprises are looking for reliability, serviceability, andaccountability (RAS) with reduced total cost of ownership torun the business.This has led to the consolidation of applications, platforms,and infrastructure. It should be noted that any consolidationexercise will primarily lead to the consolidation of underlyinginfrastructure components.The key objectives of infrastructureconsolidation include: Reduction in infrastructure footprint (locations and devices)with optimal utilization Improved reliability, availability, and serviceability ofinfrastructure by deploying enterprise-grade components Reduction in technology footprint with fewer infrastructurecomponents to manage Reduction in time to deploy new requirements Reduction in total cost of ownershipCapgemini has developed robust database consolidationmigration practices to support enterprises in their databaseconsolidation with the help of Oracle engineered systems likeSupercluster for the OS and application workloads, OracleExadata for database workload, and Oracle Exalogic formiddleware. Capgemini has set of accelerators developedto speed up migrations, including the below methodologies,tools, and processes: Infrastructure rationalization using WARP for infrastructure Application Infrastructure mapping using tools from Oracleand other industry leaders like BMC and HP Database consolidation migration Center of Excellence(CoE) comprising of application and infrastructure subjectmatter experts Reusable templates covering different phases: discover,design, define and consolidate database consolidation7

3 Cloud Deployment ArchitecturePlatform rationalization takes all layers of an IT landscapeinto consideration for rationalization, from the services andprocesses delivered to the business to networking andstorage.A best practice for platform rationalization in relation todeployment architecture is to provide as much as possible inthe form of “as a Service.” This enables enterprises to benefitthe most from standardization and provide the highest level ofagility and cost efficiency.For Oracle-centric IT landscapes, Capgemini has adopted asignificant number of Oracle best practices, to which we haveadded our own experience and proven deployment blueprints.The combined knowledge of Capgemini and as Oracleprovides a full portfolio and proven deployment architecturefor platform rationalization in an Oracle-centric IT landscape.3.1 Cloud conceptsThere are numerous definitions that are given to cloudcomputing. The most widely accepted definition of cloudcomputing is stated by the U.S. National Institute of Standardsand Technology (NIST) and describes cloud computing in thefollowing terms:Cloud computing is a model for enabling ubiquitous,convenient, on-demand network access to a shared pool ofconfigurable computing resources (e.g., networks, servers,storage, applications, and services) that can be rapidlyprovisioned and released with minimal management effort orservice provider interaction.1While within this document cloud computing will refer primarilyto the private cloud concept, it will also touch on public cloudand hybrid cloud.3.1.1Private cloud“Private cloud” refers to the model in which an enterprisecreates a cloud within the confines of a datacenter and withinits own IT landscape. In this model, the company owns thehardware assets, providing business departments and otherinternal IT departments with the benefits of cloud computing.Private cloud solutions are generally used in cases where acompany has compelling reasons to not adopt public cloud,including specific technical needs that cannot be fulfilled in thepublic cloud arena or specific regulatory and compliancy rulesand regulations.3.1.2Public cloud“Public cloud” refers to the model in which services areprovided to an end-customer from a publicly available multitenant cloud. In this model, the provider owns all the assetsof the cloud and provides computing and other services on apay-per-use model.3.1.3Hybrid cloud“Hybrid cloud” refers to the model in which an enterprisemakes use of both private cloud as well as public cloudsolutions. Often the hybrid cloud is seen as one single cloudfrom a business point of view. Business rules and workflowsallow for automated decisions to deploy applications in apublic or private cloud.1 Mell, Peter and Grance, Timothy (September 2011). The NIST Definition ofCloud Computing. NIST (National Institute of Standards and Technology)Special Publication, 800-145. Retrieved from /SP800-145.pdf8Platform Rationalization

Oracle Based Big Data Strategy the way we see itFor many enterprises, hybrid cloud is the solution of choiceconsidering it provides systems with the flexibility andscalability of cloud computing while adhering to internal andexternal regulatory and compliancy rules and regulations.3.2 Private cloud modelsa consumption-based model and internal billing can beaccommodated.The use of Oracle Enterprise Manager is included in allthe Capgemini blueprints for Oracle engineered systemsdeployments.3.2.1 Single-tenant private cloud3.3 Cloud as a Service modelsThe single-tenant private cloud is a model in which only onetenant makes use of the private cloud. “Tenant” can refer to adepartment, to a subsidiary of the owner of the private cloud,or to a partner company.Within cloud computing, a common aim among vendors,implementation parties, and enterprises is to ensure that asmuch as possible is delivered in a “as a Service” model thatallows for pay-per-use.Within the single-tenant private cloud model there is no directneed to ensure that information and services are strictlysegregated considering the owner of all the information andthe services is the same. Standard security segregation in allcases needs to be implemented based upon the company’sinternal security directives and regulatory compliancy rules.The three main “as a Service” concepts that are used withincloud computing are Infrastructure as a Service (IaaS),Platform as a Service (PaaS), and Software as a Service(SaaS).3.2.2 Multi-tenant private cloudThe multi-tenant private cloud is a model in which multipletenants make use of the private cloud. Just as for the singletenant private cloud, a department, a subsidiary of the ownerof the private cloud, or even a partner company or customerscan be tenants of this cloud.In the multi-tenant private cloud model, it is of vital importancethat all tenants are segregated strictly from each other.This requires a radically different technical and securityimplementation strategy in comparison to that associated witha single tenant private cloud.Figure 3.1: Standard cloud modelsIaaSPaaSSaaSDatabase as aServiceMiddleware as aService3.2.3 Oracle Engineered SystemsFor enterprises with an Oracle-centric deployment footprint,Capgemini recommends creating private cloud that is bothsingle- tenant and multi-tenant and is based upon Oracleengineered systems. Oracle engineered systems provide thefoundation, both in terms of hardware and software, for thecreation of a private cloud.Depending on the type of private cloud services required bythe customer, the deployment of one or more of the followingOracle engineered systems is considered valid: Oracle ExadataOracle ExalogicOracle SPARC SuperclusterOracle Private Cloud ApplianceCloud ComputingWithin the overall Capgemini strategy for platformrationalization, IaaS and PaaS are standard components thatprovide the stepping stones for an enterprise to enable privateenterprise-wide SaaS capabilities.A common model used is a combination of theaforementioned Oracle engineered systems. For managementand cloud enablement, Oracle Enterprise Manager plays avital role. Oracle Enterprise Manager provides the buildingblocks to create a private cloud including self-serviceportals, metering, and chargeback solutions to ensure that9

3.4 IaaSIaaS provides computing power in a self-service manner toend users, allowing virtual machines to be delivered directlyto an end user based upon a pre-defined template. Whenan IaaS component is requested by an end user, all relatedcomponents like storage, backup, and networking are directlyconfigured and the end user is presented with a workablevirtual machine.When deploying a private IaaS solution as part of a platformrationalization project using Capgemini architecture blueprintsand best practices, users will take advantage of the OracleEnterprise Manager self-service portals to request a new IaaScomponent. Metering and chargeback will be done via OracleEnterprise Manager.Depending on the primary goal and sizing of the enterpriseprivate cloud, new virtual machines can be deployed onOracle Exalogic, Oracle Sparc Supercluster, or Oracle PrivateCloud Appliance.3.5 PaaSPaaS provides computing power and a solution stack in aself-service manner to end users. It often makes use of IaaSto provide computing via a pre-defined solution stack. PaaSincludes solutions like middleware as a Sevice (MWaaS) anddatabase as a service (DaaS).When deploying a private PaaS solution as part of platformrationalization project using Capgemini architecture blueprintsand best practices, users will leverage the Oracle EnterpriseManager self service portal to request a new PaaS. Meteringand chargeback will be done via Oracle Enterprise Manager.Depending on the sub-components and primary goal of theprivate PaaS solution, deployment can be done with severalOracle engineered systems. For private DaaS solutions, OracleExadata is the preferred solution for deployment. For MaaS,the preferred solution is weblogic deployments on OracleExalogic.Depending on enterprise-specific needs and platform sizes,Oracle Sparc Supercluster and/or Oracle Private CloudAppliance implementations can be used as the foundation ofPaaS solutions.3.5.1 MaaSMaaS occurs when middleware is provided as a service tothe end user. With MaaS, users can request provisioning andde-provisioning of middleware including, for example, OracleWeblogic, in the same manner they would request a virtualmachine within the IaaS concept or a database in the DaaSconcept.Depending on sizing, middleware requirements, and theoverall enterprise IT landscape requirements, Capgemini10Platform Rationalizationhas blueprints for deploying a private MaaS cloud on OracleExalogic, Oracle Sparc Supercluster or the Oracle PrivateCloud appliance. For small-footprint MaaS landscapes,standard x86 Sun Oracle servers can also be used; however,in all cases Oracle Exalogic is preferred as the foundation for aprivate MaaS cloud for Oracle Weblogic Middleware.Most Fortune 1000 companies have deployed multiplemiddleware software that has overlapping functions and ispoorly integrated. We believe that most of this middlewaresoftware can be combined to remove the overlap andto enable product standardization and consolidation ofoverlapping software.For example, Oracle Weblogic has added an EnterpriseApplication Integration (EAI), functionality, thereby eliminatingthe need for separate EAI products and the differencebetween application servers and EAI products. Similarly,function activations can be done by any of the middlewaresystems like EAI, application servers, portals, etc.In the application discovery phase of Capgemini’smethodology, we identify the middleware components andmap them to the key functionalities for which they are used,like function activation, transform adaptors, messaging, andprocess flow. We then examine each of the overlappingfunctions and the reason why each of the middlewareproducts exists in the enterprise portfolio at hand. Theconclusions of this mapping are fed into the architecturecommittee to decide which of the middleware componentscan be consolidated and standardized into the finalizedmiddleware product. For example, Weblogic can be used asthe consolidated middleware layer.We recommend avoiding having more than one middlewarecomponent for webservers like IIS, Web sphere, and Apache,which can be merged into a single product easily.A pre-rationalized enterprise IT landscape is often as diverseas shown in figure 3.2.Among the characteristics of an enterprise IT landscapebefore rationalization, the following are the most noticeableand have the most negative impact on IT and businesscapabilities: Dedicated hardware and software stack per applicationHeterogeneous application componentsLack of unified monitoring and management capabilitiesComplex and costly maintenanceDifficulty determining true costsFigure 3.3 shows a post-rationalized enterprise IT landscapein which IT landscape components are reused and sharedLinux and Weblogic are implemented based upon automated

Oracle Based Big Data Strategy the way we see itFigure 3.2: Evolutionarily grown application tionApplicationJBOSSOC4JTomcatWeblogicTomcatRedhat LinuxSun SolarsWindows ServerOracle LinuxDebianHPSunDellHPDellFigure 3.3: Modern architected application landscapeOracle OptimizedApplicationOracle OptimizedApplicationOracle OptimizedApplicationOracle OptimizedApplicationOracle OptimizedApplicationOracle WeblogicOracle LinuxOracle Engineered Systems11

deployments that make use of standardized and optimizedtemplates.Among the characteristics of a post-rationalized enterprise ITlandscape the following are the most noticeable and have themost positive impact on IT and business capabilities: Shared resources Homogeneous and reusable application components Unified monitoring and management capabilities withOracle Enterprise Manager Pay-per-use internal billing options Lower complexity and reduced and more predictablemaintenance costsEnterprises need to carefully plan platform middlewareconsolidation. For example, a platform such as Oracle Fusionmiddleware has the ability to merge different middlewaretechnology into one easy-to-migrate platform. Becausethis platform can be customized for any Java application, itreduces the complexity of migrating any one product to OracleFusion Middleware. Each of the tenants gets a dedicatedOracle Weblogic cluster with built-in high availability.3.5.2 Database as a ServiceIn DBaaS, the database is provided as a service to theend user. Within the DBaaS concept, the user can requestprovisioning and de-provisioning of database in the samemanner as they would request a virtual machine for IaaS or amiddleware component for MaaS.Depending on sizing, database requirements, and overallenterprise IT landscape requirements, Capgemini hasblueprints for deploying a private DBaas cloud on OracleExadata, Oracle Sparc Supercluster, and the Oracle PrivateCloud appliance. While for small-footprint DBaaS landscapes,standard x86 Sun Oracle servers can also be used, in allcases Oracle Exadata is preferred as the foundation for aprivate DBaaS.Before rationalization takes place, the traditional enterprise ITdatabase landscape is dominated by a diverse assortmentof hardware, operating systems, and databases, all of whichrequire costly maintenance, associated support tools,procedures, and processes in addition to associated stafftraining.Among the characteristics of databases in a non-rationalizedenterprise IT landscape, the following are the most noticeableand have the greatest negative impact on IT and businesscapabilities: Dedicated hardware and software stack per databaseLack of unified monitoring and management capabilitiesComplex and costly maintenanceDifficulty determining true costs12Platform Rationalization Expensive and time-consuming deployment of newdatabases Absence of standardized database platform.Figure 3.5 shows a rationalized enterprise IT landscape fordatabases. It demonstrates how components are reused andshared. It also shows how new database implementationcan be completed based upon automated deployments thatmake use of standardized and optimized templates via OracleEnterprise Manager. The full database lifecycle is managed viaOracle Enterprise Manager and enables users to make use ofself-service portals and lower operational costs.Key advantages: Shared resources Unified monitoring and management capabilities with Oracle Enterprise ManagerLower complexityReduced and more predictable maintenance costsPay-per-use billing options are available for internal billingNew databases deployed on demand and within minutesStandardized and reliable platform for developing newdatab

2.1 Capgemini project blueprints For platform rationalization, Capgemini uses project blueprints that are standardized and industrialized to ensure rapid results and a rapid return on investment. The Capgemini WARP methodology is the foundation of Capgemini's overall approach and project blueprints. WARP is a tried-