Very few have reached that state but some companies are near.
Making Leaders Successful Every DayApril 9, 2007 Information Fabric 2.0: Enterprise Information Virtualization Gets Realby Noel Yuhanna and Mike GilpinUntitled Document 2007, Forrester Research, Inc. All rights reserved. Forrester, Forrester Wave, RoleView, Technographics, and Total Economic Impact are trademarks of Forrester Research, Inc. All other trademarks are the property of their respective companies. Forrester clients may make one attributed copy or slide of each gure contained herein. Additional reproduction is strictly prohibited. For additional reproduction rights and usage information, go to www.forrester.com. Information is based on best available resources. Opinions re ect judgment at the time and are subject to change. To purchase reprints of this document, please email firstname.lastname@example.org.T RENDS EXECUTIVE SUMMARYForrester s information fabric vision of enterprise information virtualization has evolved as enterprises have implemented more parts of the vision and as vendors have extended information-as-a service (IaaS) products to deliver broader capabilities. Although a complete information fabric implementation is still rare, many large enterprises already have some elements and a number of information services deployed or are planning deployment over the next two to three years. Although no vendor oers a complete information fabric solution, Forrester expects that by 2008, a few vendors will have su cient breadth and depth to support a comprehensive fabric solution. Information fabric 2.0 has been updated to re ect support for more business requirements, with an added focus on information quality, security, integration, and performance. Enterprise architects should evaluate Forrester s de nition of information fabric 2.0 and use it to guide their IaaS strategy.TABLE OF CONTENTSEnterprise Information Fabric Addresses A Broad Range Of Applications Information Fabric Will Become A Strategic Resource For EnterprisesInformation Fabric Architecture Is Maturing The Information Fabric Vendor Landscape Is Growing And Also ConsolidatingRECOMMENDATIONSInformation Fabric Should Guide Your IaaS Strategy Supplemental MaterialNOTES & RESOURCESForrester interviewed 36 vendor and user companies, including: BEA Systems, Cingular Wireless, Citibank, Composite Software, eBay, Endeca Technologies, GemStone Systems, GigaSpaces, IBM, Informatica, Ipedo, JPMorgan Chase, Merrill Lynch, MetaMatrix, Microsoft, Modus Operandi, Oracle, Sybase, Tangosol, Terracotta, United Parcel Service of America (UPS), Wachovia, and Xcalia,.Related Research Documents Information-As-A-Service: What s Behind This Hot New Trend? March 22, 2007, Question & Answer EDA, SOA 2.0, And Digital Business Architecture September 1, 2006, Trends Information Fabric: Enterprise Data Virtualization January 9, 2006, TrendsApril 9, 2007Information Fabric 2.0: Enterprise Information Virtualization Gets Realby Noel Yuhanna and Mike Gilpinwith Katie Smillie and Jacqueline Stone2810131819Untitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 2TARGET AUDIENCEEnterprise architect professionalENTERPRISE INFORMATION FABRIC ADDRESSES A BROAD RANGE OF APPLICATIONSTis report updates an earlier one in which Forrester introduced the concept of information fabric.1 During the course of the past year, we conducted extensive research on vendor oerings and customer case studies in the rapidly evolving market of information-as-a-service (IaaS).2 Although no vendor today oers an entire information fabric solution, many are extending their oerings to align with strategies that closely resemble Forrester s vision. Some of the key vendors that continue to lead this market are BEA Systems, IBM, and MetaMatrix, while Composite Software, GemStone Systems, Ipedo, Microsoft, and Oracle are working to oer broader and more integrated solutions. While investigating customer deployments, Forrester did not nd a complete information fabric implementation. Some large enterprises have a few elements of the fabric deployed with a goal to expand in the coming years, while others are still in the early planning stages of their implementations. However, most of the enterprises that we interviewed envisioned information fabric as an essential strategy to overcome their data integration challenges. Architects implementing a complete information fabric are often engaged with one key vendor for most of the solution, lling in gaps with either homegrown or other vendor solutions. In the course of our research, we examined customer deployments across multiple industries that use information fabric architectures to overcome integrated information management problems.Modus Operandi Helps The US Air Force Achieve Real-Time Integrated InformationModus Operandi is a software vendor that provides enterprise-level information integration for federal agencies and prime contractors. Founded in 1984, it continues to deliver information integration technologies and services to enterprises. Forrester spoke with Alan McCutchen, vice president of product development, about how Modus Operandi is helping customers like the US Air Force to support data virtualization and integration. Modus Operandi chose to partner with BEA Systems to deliver its AquaLogic Data Services Platform (ALDSP) as part of the solution.Te US Air Force was dealing with many data sources, causing data integration and real-time data access challenges and driving the need for a federated engine capable of serving a broad range of application domains. It needed to support various portlets, applications, and executive dashboards to provide an integrated view of information from multiple locations. In spring 2005, Modus Operandi helped the Air Force to deploy the Wave Platform, a model-driven data services solution using BEA s ALDSP. This solution provided the Air Force with an object-oriented view of the integrated information necessary to support its operations. As Alan stated: Untitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 3 Out of the box, the Wave Platform, combined with BEA AquaLogic Data Services Platform, provided Web services interfaces necessary to access all available enterprise data in a semantic context. This represents a signi cant paradigm shift from traditional EII [enterprise information integration] implementations on multiple levels. An important part of our solution to the Air Force was that all of the data integration and virtualization was to be machine-consumable, not just applications or end users, [and] therefore had to be high performance, scalable, automated, secure, and seamlessly integrated. Te solution enabled the Air Force to integrate semistructured and unstructured document archives. ALDSP was able to integrate data in real time from many dierent data sources, which was essential in providing stakeholders with relevant, integrated, and real-time information. Although ALDSP was a critical component of the solution, Modus Operandi also had to build custom extensions, especially in the areas of abstraction and support for OWL Web Ontology Language ontologies. Alan told us: The BEA solution did an excellent job at the schema level for both logical and physical design, along with tapping various metadata repositories and information, but did not excel at the semantic layers of abstraction transformation, aggregation, and rollup, which was the missing piece. We were able to customize the solution to ll in the gaps to meet customer requirements. Merrill Lynch Uses MetaMatrix To Achieve A Uni ed View of InformationMerrill Lynch is a leading nancial services company with total client assets of more than 1.5 trillion and o ces in 37 countries worldwide. It is one of the leading global underwriters of debt and equity securities and is a strategic advisor to corporations, governments, institutions, and individuals. Like the Air Force, Merrill Lynch faced challenges with multiple data sources and large amounts of data. It wanted a solution that could create a single view across multiple data sources, cut the amount of time required to integrate new applications with databases, and provide real-time information access to meet growing new requirements. The company identi ed the need for a standard, global interface to all of its data sources beginning with core enterprise data in ve categories: product, pricing, party, organizational, and corporate actions. Merrill Lynch chose MetaMatrix to help create a uni ed view of information stored in DB2, Oracle, and Sybase systems. The key bene ts cited were increased operational e ciency, reduced development time and eort, and a more accurate view of enterprise data, all of which helped business users make faster decisions. Previously, developers had to write custom programs to pull data from various data sources, which created a challenge for information managers to keep track of all of the dierent ways information was being accessed and used. According to Merrill Untitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 4Lynch, Using MetaMatrix has cut the amount of time needed to integrate new applications with the databases and reduced the integration workload for the programmers. MetaMatrix provides a single view of enterprise data to portals, dashboards, desktop applications, and traditional applications. It also has helped Merrill Lynch provide information more quickly to end users to help them make faster business decisions.Wachovia Chooses Tangosol To Meet High-Performance Application RequirementsWachovia is one of the largest nancial institutions in the US, operating in 21 states and with more than 40 o ces around the world. It is highly diversi ed, providing a broad range of banking, asset management, wealth management, and corporate and investment banking products and services. Forrester spoke with Dipen Mehta, vice president of utility computing - corporate and investment banking technology, to understand Wachovia s strategy when dealing with high-performance applications.Wachovia was suering from the data-access demands of a high and growing volume of credit risk transactions, and it concluded that distributed caching technology was the answer. It evaluated several o-the-shelf caching solutions and chose Tangosol as the best t for the requirements of credit risk processing. This application needed to process hundreds of gigabytes of information with low latency, so architects used Tangosol to spread this data across 150 cache nodes in a larger grid of more than 6,800 compute nodes, managed with technology from DataSynapse. This not only spread out the data-access workload but also increased the total computing power available to the credit risk application. The credit risk application stores more than 200 million objects in this cache, and Wachovia has not yet encountered any major limitations, even though objects can reach up to 3.5 megabytes in size. Dipen told us: Tangosol maintains a single state image every node always knows about every other node. Each node knows whether a particular object is on that node, and this is true whether it s ve or a thousand nodes. No matter how large a cluster is, there s always a single owning node of a particular piece of data. Access to that data is serialized, so the integrity of the data is assured. The scalability and performance of Tangosol works as advertised. In fact, we had to tune Windows frame and MTU [maximum transaction unit] sizes to get network performance to live up to the potential of Tangosol. One important lesson from the credit risk implementation is the importance of designing the application while keeping in mind the dynamics of data access within the cache. Wachovia ensured that each cached object was most likely to be accessed by application code residing on the same node, minimizing network chatter to sync the cache. Each application object ensured integrity, as well as performance, by providing the serialized access interfaces for managing locally cached data. Future information fabrics will automate more of this optimization, reducing application design complexity.Untitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 5A Manufacturing Company Applies IBM s Information Server To Enhance Business DecisionsForrester spoke with a manager of enterprise resource planning (ERP) solutions at a large manufacturing company in North America to understand how the company deals with data integration and information management challenges. According to the manager, the company has a number of heterogeneous databases, including DB2, Oracle, and SQL Server, that support a wide range of information across the business. This includes information about business performance, customer analytics, nance, product lines, and key company metrics. These databases run on various IBM AIX and Windows servers across several data centers. The manager told us: Lots of business projects are driving the demand for information. Our system landscape is extremely heterogeneous, so we are tapping more than 50 dierent data sources. They are all very heterogeneous from a database perspective, besides ftp les, at les, Excel, XML, you name it. A key challenge for the company is to ensure that the correct information is supplied to applications, with su cient speed. In addition, the company needs to support business intelligence requirements. Te manager stated: Prior to using IBM s Information Server components, we had nothing comparable. The technologies we had were whatever the particular application provided, or else we had to write program code to get the data. Te manufacturing company is a longtime customer of IBM and has used several information server components even before they were bundled into IBM s new oering. These included dierent editions of DataStage, Pro leStage, QualityStage, and SOA/RTI. The manager indicated that Information Server helps integrate these and other components to deliver higher business value. Since using the IBM solution, the manufacturing company was able to develop 22 analytical applications in areas including customer management, dealer nancial performance analytics, employee analytics, corporate safety, product diagnostic analysis, labor and time reporting, product costing, product line pro tability, parts and accessories, warranty claim analysis, rental product operations, dealer/call-center single point of contact, pricing analysis, and retail pricing analysis. According to the manager: The key requirement for IBM s Information Server was the business demand for business intelligence, delivering the right information at the right time. We wanted the most cost-eective solution to meet functional requirements, besides features and functionality, and IBM Information Server delivered that. We also found it s best not to propagate data more widely than necessary, keeping it organized and cleansed and controlling the growing Untitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 6volume. We used the Information Server middleware to support common information requirements on a common integrated platform, rather than building a custom solution. Tis has helped us reduce cost. A Year Of Experience Teaches Many Information Fabric LessonsInformation fabric deployment continues to grow, but this growth is concentrated in the nancial services, telecommunications, retail, and government sectors. Most enterprises deploying information fabric are large companies with more than 1 billion in revenue, although a few midsize companies are also participating. Since most of these projects take up to three or four years to reach all of their goals, incremental implementation is essential. A common lesson from several deployments is that a complete information fabric deployment isn t necessary to begin receiving bene ts; bene ts begin to accrue even when only a few fabric elements are in place. For example, architects can deploy fabric technologies such as distributed caching, federated queries, metadata repositories, and enterprise search functionality independently of each other. Additional lessons learned during the past 12 months include: Most usage of the fabric is read-only. Although more than 90% of current information fabric deployments are read-only, several architects are planning to add write capability in the near future. Forrester expects that by 2010, one-third of information fabric deployments will support both read and write capabilities. Most fabrics begin as read-only and add write capability only later (if at all), after establishing reliability and accuracy. The most common update pattern today is to read through the fabric but update by sending a service request (message) to a transactional business service that updates the underlying system of record. Additional components are needed to support an enterprisewide fabric. Although the logical view of the information fabric architecture has not changed, Forrester now views additional fabric components, such as information quality, transformation, and search, as key requirements for a comprehensive deployment (see Figure 1). Since the information access layer taps disparate and redundant data sources that may disagree, architects are likely to nd that ensuring information quality is a critical requirement. Moreover, data is often stored in dierent formats in dierent applications and databases, so architects are also likely to nd it necessary to transform the data to meet the needs of new information consumers. Thus, information quality and transformation are critical components for many successful information fabric deployments. No single vendor oers a complete information fabric solution. Although vendors have made considerable progress in the past year in oering more features and functionality, no single vendor has a complete solution yet, although in partnership, they can sometimes get close. BEA, IBM, and MetaMatrix still lead the pack, with strong potential shown by Composite Software, GemStone, Microsoft, Oracle, Sybase, and Xcalia.Untitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 7Figure 1 Logical Design Large enterprises are planning to deploy complete information fabric solutions. Many large enterprises, especially in the nancial services, telecommunications, and government sectors, are focusing on enterprisewide information fabric deployment. Most of these customers are either looking to support a single version of the truth or improve business decisions through portals and executive dashboards. Forrester expects that one-third of all enterprises with revenues of more than 1 billion will have an information fabric architecture by 2012 as part of an overall IaaS strategy. Both design-time and runtime metadata repositories are important. Information fabric is not just about delivering data to applications at runtime; it is also about helping developers and architects to develop new applications more easily. The design-time metadata repository helps developers to more easily discover and consume information services to deliver data through the fabric for the new applications they are building. Performance and scalability are still critical barriers. When applications decouple from the data to support information fabric architecture, a key concern that many architects have is the impact on the application s response time. When companies are using high-performing applications or dealing with large amounts of data, Forrester highly recommends that architects put special emphasis on distributed data caching capabilities to ensure acceptable response time to consumers. A persistent cache or other intermediate store may also be needed when the resource cost to reconstitute the intermediate representation is high compared with the cost of redundancy and synchronization.r Frr rR rhIn1 7BackupsDatawarehouses OtherfabricInfrastructureApplications Grids DatabasesFilesInfo fabricInfo sourcesInfomgtVirtual view of enterprise informationIntegrationCatalogSecurityDiscoveryEventsManagementTransformationQualityMetadataApplications PortalsEnd users Other fabricUsers & appsUntitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 8 Information consumers and architects are concerned about data quality. As one architect stated: I don t know where the data is coming from when it comes through the fabric. My rst question is How accurate and reliable is the data provided through the middleware? Architects should verify that all semantic mappings and other metadata are thoroughly tested and that ongoing procedures for maintaining the fabric metadata will be rigorous in protecting quality. If the information provided by the fabric is to become a trusted critical resource, architects should make data quality assurance and reliability testing a high priority. Information users need a friendly interface to the fabric. Although information fabric 1.0 primarily focused on applications, Forrester found increasing demand for the ability to dynamically discover, search, and browse the newly integrated information provided by the fabric in real time. Such searches enable information users to address business requirements, such as compliance and audit reporting, business intelligence, and analytics.INFORMATION FABRIC WILL BECOME A STRATEGIC RESOURCE FOR ENTERPRISESAlthough information fabric 1.0 put a special emphasis on applications that needed a single version of the truth for enterprise information, our research found that IaaS is also supporting other uses, such as transaction processing, decision support, business intelligence, content management, B2B, and radio frequency identi cation (RFID). In addition to supporting applications, the fabric can also be used by end users using a Web browser to search and browse all types of enterprise information. An information fabric can provide a 360-degree view of all enterprise information, not just customer or nancial data, although architects should work incrementally toward this goal. Forrester expects that information fabric will become a strategic resource and the primary source of information for new applications over the next seven to 10 years. Some of the key consumers of information fabric include: Applications that need a single version of the truth for enterprise information. Many enterprises struggle to deliver accurate data to applications in real time because many deal with hundreds and thousands of databases and large amounts of unstructured data, making it hard to nd a single version of the truth. Information fabric overcomes this challenge by focusing on a virtual view of cleansed enterprise information to deliver consistent and reliable data to support all types of applications and end user requirements. Business intelligence and dashboards. In addition to supporting traditional applications, information fabric also oers greater exibility in delivering aggregated information for better business decisions. These decisions are supported through business dashboards, metrics, and reporting. As information ows through the fabric, it provides data cleansing and also oers the ability to perform aggregation, summarization, and transformation. Forrester found that several large enterprises are using information fabric architecture to support industry-speci c metrics, nancial management and reporting, asset and liability management, and risk management. Untitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 9 Compliance and audit reporting. Nearly all enterprises are feeling increased pressure to deliver regular reports to meet compliance requirements, as well as real-time business information to support auditors requirements. Information fabric helps enterprises get information quickly and accurately to meet such dynamic requests, which may be aggregated, transformed, or accessed from several sources seamlessly. An important feature of advanced fabric implementations, and one that is critical to enabling its use for auditing, is the ability to show the complete provenance of any information the fabric provides, citing original sources, extraction time, and transformations. Searching and browsing of enterprise data. Searching enterprise information is usually not as straightforward as performing an Internet search. This is because enterprise data is stored across hundreds of disparate databases and les and must deal with both structured and unstructured data. An information fabric can help developers or even users access any type of information more easily without having to know the data structure or other metadata information or its location. Forrester believes that searching and browsing virtualized data in real time will ultimately become the most widely used application of information fabric. Content management applications. As many of today s leading content repositories have added industry-standard Web services interfaces, it is no longer necessary to employ proprietary technology that is unique to each vendor to obtain content.3 Information services that need to combine content from multiple repositories from dierent vendors can now use industry-standard protocols to access and aggregate this content in the fabric. This is often part of a master data solution that bridges structured data from databases to this aggregated content. A common pattern for customer information services is to provide cleansed, master customer data combined with content such as contracts, faxes, or photographs (i.e., pictures of a damaged car to support an insurance claim). Content is usually linked by references that point to the content repository, rather than by embedding the content directly into the merged customer information record. Other fabric. Forrester expects that in the future, enterprises will have multiple information fabrics deployed, catering to dierent business domains. To support uni ed information across the enterprise, a consumer of information fabric can also be another fabric: a exible, federated architecture that can expand the scope of information availability (see Figure 2). Over the next two to three years, Forrester expects that architects designing integration with business partners and other enterprises will begin to make use of IaaS and fabrics, creating an even more exible architecture for information sharing and delivery.Untitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 10Figure 2 Multiple-Fabric IntegrationINFORMATION FABRIC ARCHITECTURE IS MATURINGBased on our interviews of customers and vendors over the past year, Forrester s de nition of the architecture of an information fabric has evolved to support more applications and business requirements. From the original four components in information fabric 1.0, there are now 10 key technology elements that comprise information fabric 2.0 (see Figure 3). Forrester expects information fabric technology to provide industry-standard interfaces, including XML, SOAP, ODBC, Java Database Connectivity (JDBC), and HTTP, to support all types of applications and end user requirements. Besides the 10 key technologies needed within the fabric, it will also be critical for architects to ensure that monitoring and management infrastructure can also manage the fabric. Tis requires holistic monitoring of all relevant servers, applications, databases, and middleware. To facilitate holistic monitoring, fabric components should interface with industry-leading management solutions, such as CA s Unicenter, HP s OpenView, and IBM s Tivoli. These are the 10 key elements of information fabric 2.0: Service registry and discovery (new). Application integration with the fabric can be done either by incorporating services at design time or by having the application automatically discover services at runtime. Most developers today are binding application data to information fabric services at design time, but many are looking at auto-discovery of services as a future requirement, especially to support search and browse or dynamic applications. A UDDI service r F rrr Rr h In1387Enterprise AEnterprise BPartner AUntitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 11registry acts as a directory or catalog that provides an index of services available to handle applications information requests. For example, a dynamic application might require accurate information pertaining to a customer s address; therefore, the application could search the service registry for address information and retrieve the address via the service it discovers. In addition, end users can directly browse or retrieve information related to services using the service registry. Distributed information cache. One key concern that most architects face when dealing with information fabric is delivery response time. As one architect pointed out, If middleware takes on the information management role, it needs to guarantee reliable and fast delivery. Applications traditionally have been tightly coupled with data sources such as databases and often are highly optimized to deliver fast response time. With a middle-tier data access layer, response time is bound to be aected because of the latency added by retrieving data from disparate sources. To overcome this, a distributed data cache becomes a critical component. Forrester expects that over the next ve years, large, mission-critical applications are likely to use a distributed data cache as the initial tier for all data access, with traditional databases as the persistent store.Figure 3 Physical View Of Information Fabricr Frr rR rhIn1 7Distributed information cache ApplicationsUnstructured data Structured data Semistructured dataSearchdashboardsIT/industry standardsBusiness taxonomyDesign-time metadata repositoryIntegrated enterprise search engineDistributed information accessOther fabricPortalsService registry and discoveryDistributed runtime metadata repositoryIT policiesBusiness rulesSOAP SQLJDBC HTTPXMLODBCRFIDB2B appsComplianceBusiness intelligenceDecision-support appsTransactional appsWeb servicesContent mgt appsFiles OtherfabricDevices Datawarehouses DatabasesInformation fabric Untitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 12 Design-time metadata repository (new). Applications that are being written to work with information fabric need to know what services to invoke. For this, developers must interface with a design-time metadata repository and service registry to ensure that appropriate service requests are made. The design-time metadata repository also provides semantic mapping to integrate disparate information sources, as it is the locus of the canonical information model and the mapping from that model to original information sources.4 Distributed runtime metadata repository (new). A runtime metadata repository is similar to the design-time metadata repository, except that it also contains information about the location of data and its state (if cached data is consistent with the back-end persistent data store or if metadata guides the sync of remote caches). The runtime metadata repository provides semantic mapping, the access path, and other metadata pertaining to aggregation, rollup, and summarization. One key customer observation regarding metadata repositories is that since the industry still lacks a comprehensive standard to tie repositories together, architects have no choice but to deal with proprietary, vendor-driven metadata repositories for now. Integrated enterprise search engine (new). A frequent requirement in most enterprises is the ability to search or browse enterprise information in real time. For example, an auditor may request company information from all of the sources available in an organization. Such requests are dynamic and real-time and would, therefore, require a search engine to interact with various metadata repositories and provide the relevant results, whether the data is structured, unstructured, or semistructured. An information fabric search engine provides seamless access to all enterprise data, just as Google delivers information on the Internet. Distributed information access. Tis component provides access to disparate data from various sources, including databases, les, devices, backups, data warehouses, and even other fabrics. The critical requirement is to support all types of data and be able access them quickly and consistently. The more widely distributed and diverse the underlying information sources, the more likely that it will be a requirement for the information access layer to implement advanced distributed query optimization, driven by dynamic metadata and network behavior. Transaction management (new). Two-phase commit technology supports transactions across multiple databases, but has several limitations: 1) It typically does not support multiple database management system (DBMS) products in the same transaction; 2) it requires a transaction coordinator and all the resource managers to be available throughout the transaction; and 3) it often consumes large quantities of server and network resources. The XA protocol can overcome some of these limitations, such as support for both synchronous and asynchronous transactions across distributed databases, but it still suers from many others, such as a lack of support for unstructured information. Untitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 13In the future, support for the WS-Transaction speci cation may bring richer transaction capabilities for Web-services-based information services, but today, no industry standard has been widely deployed to support granular transactions for an information fabric. Therefore, where fabric implementations provide more advanced support for transaction models, such as compensation, architects must use proprietary approaches that today work only within the con nes of one vendor s fabric implementation. Information quality (new). Since information fabric can integrate with all types of disparate data sources, information quality becomes a top priority. Data might require cleaning, ltering, and consistency checking to ensure high quality, and this has to be done in real time. None of the pure-play IaaS vendors have a strong focus on information quality in an information fabric, but many vendors are putting more focus on this area for upcoming releases. Therefore, information quality features will become a common capability of information fabric solutions over the next two to three years. Transformation (new). Te consumers of the information fabric can be any type of application, an end user, a portal, or other fabrics. Each consumer might require data to be received in a particular format and size; therefore, translation and transformation becomes an important requirement. For example, one application might want dates provided according to US convention (month, day, year), while another wants the European format (day, month, year). Information security (new). An important component of the information fabric is data access security. The information fabric architecture enables centralized data access and control, enforcing an even stricter level of data access security than what is found in traditional applications and DBMS. With the information fabric, each person s access to information is based on con rmed identity in a speci c usage context, oering a more granular level of access control. Therefore, when designing an information fabric strategy, architects and security professionals should make data access security a top priority, right from the initial planning stage. THE INFORMATION FABRIC VENDOR LANDSCAPE IS GROWING AND ALSO CONSOLIDATINGAlthough several leading information fabric vendors have made considerable progress with new features and functionality, no single vendor currently oers a complete integrated information fabric solution. BEA, IBM, and MetaMatrix continue to exhibit technology leadership of the information fabric market by having the largest number of integrated fabric elements (see Figure 4). Composite Software, GemStone, Oracle, Microsoft, and Sybase are also ramping up their solutions, extending features, and driving more integration within their product oerings. Forrester expects that by 2009, some vendors will oer a complete integrated information fabric, with some oering complete solutions in 2008 that are not yet fully integrated. Key vendors of information fabric include:Untitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 14Figure 4 Vendor Landscape BEA Systems. Recently, BEA announced its Q4 2006 earnings, and its AquaLogic revenue grew more rapidly than other BEA product lines and now represents 25% of the rm s license revenue. Tis indicates strong momentum in the service-oriented architecture (SOA) infrastructure and data services space for BEA, although a good part of this growth was driven by its business process management (BPM) product. Except for its lack of support for comprehensive data quality and distributed data caching components, BEA remains a strong and innovative player in the information fabric space. It already has several large customer deployments in the nancial services, insurance, government, and manufacturing sectors. ALDSP is capable of supporting heterogeneous data sources, including support for packaged applications such as SAP. The recent release of ALDSP 2.5 oers fast path integration with AquaLogic Service Bus, add-in support for Microsoft Excel, and a native SQL engine that can also access the fabric in real time.Source: Forrester Research Inc.1 7Distributed information accessInformation qualityDistributed transaction managementInformation transformation and assemblingInformation securityDistributed runtime metadata repositoryIntegrated enterprise searchService registry and discoveryDesign-time metadata repositoryYesPartial= PartnerDistributed information cache PPPPNote: A blank cell indicates that the vendor s product does not support the functionality as it relates to IaaS.Untitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 15 Composite Sofware. Although Composite Sofware also began as an EII player and is still positioned in that way, it oers several information fabric components, including service registry, caching, metadata repository, distributed information access, security, and distributed transaction management. It focuses on leveraging existing data assets to enable SOA and is likely to have a more visible position in the information fabric market in the future. Composite has several large customers in the nancial services, insurance, telecommunications, and government sectors. The company oers integration of its metadata repositories with packaged applications, such as Oracle (including Siebel), salesforce.com, and SAP. Customers using Composite claim that information access is often fast, easy to use, and secure. However, Composite still lags in addressing information fabric architecture in the areas of data quality and enterprise search. GemStone Systems. GemStone oers a middle-tier data layer that supports data virtualization across disparate data sources and data types. The rm focuses on distributed resource management, in-memory distributed caching, object management, and event processing. GemFire is a suite of products that deliver a distributed data infrastructure to enable real-time delivery and access. GemStone has strong support for distributed data caching, performance and scalability, and access to information using C++, SOAP, XML, C#, and SQL. However, it lacks comprehensive support for a metadata repository, unstructured data, enterprise search, data quality, transformation, and security components to support information fabric architecture. GigaSpaces. Focused primarily on high-performance transactional and real-time analytical applications, GigaSpaces provides a middleware-virtualized layer with clustering, caching, and high-availability functionality for Java/J2EE, .NET, and C++ applications. GigaSpaces supports data grids via JDBC (SQL), Map/JCache interfaces, and JavaSpaces. Common caching topologies supported by GigaSpaces data grid architecture include replication of the data. GigaSpaces has customers in various verticals, including nancial services, telecommunications, and government. Although the company has strong support for transactions caching, it still lacks comprehensive support for data quality, enterprise search, and design-time metadata repository. IBM. During the second half of 2006, IBM made a strong move in the IaaS market by announcing the IBM Information Server, which combines several existing technologies and leverages shared metadata across products.5 IBM is one of the leading vendors in the information fabric space, but it still needs to do more to integrate all of its various fabric components seamlessly, which is likely to take two to three years to complete. Informatica. Positioned primarily as a pure-play data integration vendor that focuses on integration of disparate data sources, Informatica also oers an extract, transform, and load (ETL) solution. It has strong support for data quality, transformation, and movement, but when Untitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 16it comes to information fabric, it still lags in distributed data caching transaction management, security management, service registry, and shared metadata repository. Informatica purchased Composite Software s EII source code and oers a data federation option based on this code, which is tightly integrated with the PowerCenter product, providing a single, uni ed enterprise data integration platform. Ipedo. Founded in 2000, the company primarily focuses on EII. It oers access using common protocols, including SQL, SOAP, UDDI, JDBC, ODBC, XML, XQUERY, and WSDL. It provides built-in aggregate computation on Web services and has a built-in XML rules engine. The recent Ipedo XIP 4.2 release focuses on data governance, Web services publishing, and dynamic failover redirection. Ipedo oers support for most packaged applications, including salesforce.com, SAP, and Siebel, as well as heterogeneous DBMSes. It supports modeling tools to build logical and physical views. The rm supports metadata repositories, caching, and transaction management but lags in data quality, transformation, and enterprise search as it pertains to information fabric. MetaMatrix. Previously an EII player, MetaMatrix is now one of the technology leaders in information fabric, along with IBM and BEA. Although it has comprehensive support for information fabric components, it lacks support for enterprise search and data quality. MetaMatrix has credible customer deployments that span various vertical industries, including nancial services, telecommunications, government, and retail. In August 2006, MetaMatrix further extended its data services oering by making enhancements to include XQuery transformation, Web services metadata import, XSD message visualization, and performance improvements. In addition, by partnering with DataDirect Technologies, it now extends its data source coverage to mainframe data, including Adabas, Natural, CICS, DB2, IMS/DB, and VSAM. Microsof. Microsof has support for a service registry, design-time metadata repository, search, data quality and transformation, and transaction and security management, yet it still trails behind IBM and Oracle in the information fabric space. Microsoft has not yet integrated all of its information fabric components and therefore requires architects and developers to do the integration themselves. In addition, Microsoft does not oer strong support for distributed data caching or federated data access across dierent platforms and databases. However, the company is developing a new technology that will be relevant to information services: the Entity Data Model (EDM). EDM is part of a future release of ADO.NET, which is purely for programmers but will provide developers with an abstraction similar to some information service use cases. Oracle. Oracle s focus on IaaS has been average, mainly concentrating on homogeneous data like Oracle-related applications and databases. However, it does oer many of the information fabric components, including metadata repository, caching, search, data quality Untitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 17and transformation, security, and transactional management. These are not oered as an IaaS integrated solution, although Oracle Consulting has done some projects to integrate these components for individual customers doing IaaS. Recently, the rm announced an agreement to acquire Tangosol, which puts it ahead of other DBMS vendors in distributed data caching and in-memory technologies and one step closer in oering a complete IaaS oering. Forrester expects Oracle to integrate these components in the near future, and it could oer a complete IaaS solution as early as 2008. Progress Sofware. Progress Sofware is another player that has some of the key components needed to assemble an information fabric solution, but it has not yet done so. Its capabilities are spread across many product lines: DataDirect, DataXtend, Sonic Software, and others. Some of Progress customers have integrated its technologies to implement information services within a broader IaaS strategy, but Progress does not yet market these elements combined in this way. Forrester expects it to do so, albeit gradually and with a conservative business approach. Sybase. Nearly two years afer the acquisition of Avaki, a company that focused on enterprise data virtualization, Sybase has not aggressively exploited its market potential. The company does have several key information fabric components, including metadata repository, service registry, data quality and transformation, transaction management, and enterprise search, but it trails in security management and runtime metadata repository as it relates to information fabric. Forrester expects to see Sybase integrating its information fabric components over the next two years. Tangosol. Tis company focuses primarily on in-memory data management and caching solutions for J2EE and .NET applications. Tangosol virtualizes data and services, which it performs by coordinating updates of data using clusterwide concurrency control. This is an advanced approach, but it doesn t work equally well for all customer use cases requiring caching. Tangosol supports application servers such as Apache Tomcat, BEA WebLogic, IBM WebSphere, and JBoss. Although its distributed caching solution is among the most advanced, it lags behind other information fabric players in support for service registry, enterprise search, data quality and transformation, security management, and transaction management to support data virtualization. Combined with other vendors information fabric products, Tangosol can provide a good solution when the partner product supplies these missing elements but requires a caching partner. Recently, Oracle made a bid to acquire Tangosol. This is a win-win situation for both companies that will help take in-memory and distributed data caching technologies to the mainstream, especially when implementing IaaS. Terracotta. Terracotta oers a distributed data cache to deliver high-performance throughput for large Java applications, but its main business focus is on the use of its Java Virtual Machine (JVM) clustering technology to enable Java applications to scale more easily. Terracotta currently does not support a global metadata repository or data access across heterogeneous Untitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 18platforms. Recently, it moved to an open source licensing model for its Java clustering product in an eort to increase adoption. Xcalia. Founded in Paris in 2000 to provide data access middleware, Xcalia now provides a data services solution for building composite applications using SOA. Xcalia has more than 50 customers across several vertical industries, including insurance, banking, retail, and telecommunications. Its solutions are primarily focused on meeting the needs of developers building new applications that need to integrate access to a wide range of disparate data sources and application services. Its earlier solutions were more comparable to a Java object/relational mapping technology, but since mid-2006, Xcalia has added support for information services. Although the rm has strong support for several fabric components, including distributed transaction management, distributed information access, and transformation, it does not oer data quality and information security. The solution features the Xcalia Intermediation Core (XIC) as an enterprise-service-bus-like foundation that includes dynamic business process orchestration and a metadata repository.R E C O M M E N D A T I O N SINFORMATION FABRIC SHOULD GUIDE YOUR IAAS STRATEGYImplementing the full information fabric architecture remains complex, and not all enterprises require all components; therefore, architects should focus on a few components initially and build the architecture over time. Only architects facing the most advanced requirements should consider a broader implementation in the near term. To start building the information fabric, enterprises should keep these key points in mind: Look at vendors that oer comprehensive information fabric solutions. Architects who want to take a fast path to a more complete information fabric architecture should begin with vendors that oer the broadest set of components, such as BEA, IBM, and MetaMatrix. Expect customization and integration eorts. With no single vendor oering a complete and integrated information fabric solution, architects and developers will initially need to expend additional eorts to ll gaps and integrate best-of-breed add-ons. Don t ignore information quality. When architecting information fabric, data quality is an important component that sometimes takes a back seat. Architects should ensure that data quality is given top-priority rights from the initial planning stages. Existing applications will usually require changes to integrate with the fabric. For existing applications to invoke information fabric services, architects should plan for some application customization, except where the fabric happens to provide the style of interface already in use, such as SQL. Architects and developers should usually plan to begin with a read-only interface to the fabric and extend to include write capability after proving the Untitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 19initial architecture. Another factor that can drive the need for application change is design optimization for cache usage, as in the Wachovia example. Distributed data caching remains critical. Since accessing disparate data sources in real time is often a slow process, rms should consider distributed data caching to deliver fast response times, along with other options like staging information to intermediate storage managed by the fabric.SUPPLEMENTAL MATERIALCompanies Interviewed For This Document BEA SystemsCingular WirelessCitibankComposite SoftwareeBayEndeca TechnologiesGemStone SystemsGigaSpacesIBMInformaticaIpedoJP Morgan ChaseMetaMatrixMicrosoftModus OperandiOracleSybaseTangosolTerracottaUnited Parcel Service of America (UPS)WachoviaXcaliaENDNOTES1 Enterprises are facing growing challenges in using disparate sources of data managed by dierent applications, including problems with integration, security, performance, availability, and quality. Business users need fast, real-time, and reliable information to make business decisions, while IT wants to lower costs, minimize complexity, and improve operational e ciency. New technology is emerging that Forrester has dubbed information fabric, de ned as a virtualized data layer that integrates heterogeneous data and content repositories in real time. See the January 9, 2006, Trends Information Fabric: Enterprise Data Virtualization. 2 IaaS has exploded on the scene over the past two years, moving from an obscure topic to one of the top usage scenarios in SOA. Forrester expects that in 2007, a majority of large enterprises will add SOA to the list of ways they integrate information. See the March 22, 2007, Question & Answer Information-As-A-Service: What s Behind This Hot New Trend? Untitled DocumentTrends | Information Fabric 2.0: Enterprise Information Virtualization Gets Real 2007, Forrester Research, Inc. Reproduction ProhibitedApril 9, 2007 203 As an example of this trend, consider Microsoft s enterprise content management (ECM) support in O ce SharePoint Server 2007. The platform consists of a common content storage layer based on Windows SharePoint Services a core set of ECM services powered by O ce SharePoint Server 2007 and speci c ECM Suite components addressing document management, records management, Web content management, and eForms management. It oers user interface options within the Microsoft O ce 2007 client applications and O ce SharePoint Server 2007 and to third-party applications via Web services. See the March 2, 2007, Tech Choices Microsoft s 2007 Enterprise Content Management Platform. 4 Although a canonical model is not an essential feature of an information fabric or an IaaS strategy, it is a common pattern in more mature organizations using IaaS. See the March 22, 2007, Question & Answer Information-As-A-Service: What s Behind This Hot New Trend? 5 IBM s Information Server puts IBM ahead of its competition in total functionality, but integration of its various components will still take time and eort. See the December 28, 2006, Quick Take Information Server Takes IBM One Step Close To Delivering Information-As-A-Service. Untitled DocumentForrester Research, Inc. (NASDAQ: FORR) is an independent technology and market research company that provides pragmatic and forward-thinking advice to global leaders in business and technology. For more than 23 years, Forrester has been making leaders successful every day through its proprietary research, consulting, events, and peer-to-peer executive programs. For more information, visit www.forrester.com.AustraliaBrazilCanadaDenmarkFranceGermanyHong KongIndiaIsraelJapanKoreaThe NetherlandsSwitzerlandUnited KingdomUnited StatesHeadquartersForrester Research, Inc.400 Technology SquareCambridge, MA 02139 USATel: +1 617/613-6000 Fax: +1 617/613-5000Email: email@example.comNasdaq symbol: FORRwww.forrester.comM a k i n g L e a d e r s S u c c e s s f u l E v e r y D a yFor a complete list of worldwide locations,visit www.forrester.com/about.Research and Sales O ces41387For information on hard-copy or electronic reprints, please contact the Client Resource Center at +1 866/367-7378, +1 617/617-5730, or firstname.lastname@example.org. We oer quantity discounts and special pricing for academic and nonpro t institutions.