Sun Gets Head Start on SOA Registry
Type: Competitive Intelligence Report
Analyst: S. Willett
Report Date: June 21, 2005
Module: Application Infrastructure
ID: CIR14024 |
Current Perspective: Positive/Neutral
Vendor Importance: Moderate
Market Impact: High |
|
Summary
Event Summary
June 15, 2005 — Sun Microsystems, Inc. (NASDAQ: SUNW) today announced the early access availability of the Sun Service Registry for customers who need to track and manage increasing numbers of Web services. Sun's Service Registry also includes an integrated repository for storing service metadata and providing additional capabilities, such as Web services lifecycle management. The combined registry-repository serves as a cornerstone for true service oriented architecture (SOA) governance.
Analytical Summary
• Current Perspective: Slightly positive on Sun’s Services Registry, as the company is showing vision in coming up with an SOA management and control platform, although it needs to flesh out all the elements and ensure the ebXML 3.0 repository becomes a standard.
• Vendor Importance: Moderate to Sun, as the firm needed to come up with a more neutral SOA development/management platform to attract users to its JES middleware products set.
• Market Impact: High on the SOA integration/management market, as competitors who want to gain a stake in this segment must either include this technology or partner for it, to meet IT governance requirements.
Top
Perspective
Current Perspective: Positive/Neutral
We are taking a slightly positive stance on Sun’s new Services Registry. Sun is showing some vision here in propagating a combined UDDI/ebXML 3.0, which could be the foundation for managing and even controlling services in large SOA environments. The registry combines UDDI 3.0 registry (with basic discovery) with an ebXML 3.0 repository. In addition to being able to store all kinds of metadata and “artifacts” about services (transformations, XML schema descriptions, BPEL, etc.), ebXML can be “federated” with other ebXML repositories. This means that users can do queries that span registries from different vendors (as long as they are ebXML 3.0-compliant) and provide management of services in an extended network. Sun is building out a number of SOA governance features in its product. For example, it has an access control feature and one for content-based event notification feature, as well as a feature to version services. Users can also define taxonomies and construct associations and classifications of services. Other planned features, such as cataloguing and validation, will be accommodated through a plug-in. While not all of these features are fully developed with UI and templates (the first version of the product is, after all, not even in beta), the range of features will help administrators get a rough handle on their SOA and do some basic management. It puts them ahead, at least at this point, with competitors such as IBM and BEA.
Sun plans to leverage its registry to gain an advantage for its Java Enterprise System (JES) middleware platform. The company will build in integration to the Registry from the firm’s security, portal, and development tools over the next year or so. The Service Registry could be a platform for all types of ISVs and could tie into ISV applications. This can include management vendors and security vendors who want to put policies and QoS type of rules in a central place. It can also include application vendors.
Sun, however, has not yet shipped the beta version of the product, and many features, such as UI, have not been completed. This will give competitors plenty of time to respond. For example, IBM, Microsoft, and BEA (through Systinet) have shipped out UDDI 3.0 directories, but have not yet enhanced the interfaces or linked these into other middleware and management products. Expect them to do so in the next 12 months. Sun is also gambling that ebXML 3.0 repository will gain credence as a standard. This is not unreasonable, since it has a lot of advantages and is widely available through the open source world. However, IBM and Microsoft have not yet staked out a position, and they could just opt for further development of UDDI. Sun’s ISV program has also not yet matured. The firm should consider seeding the market with its registry, especially to integration, systems management, SOA management, and application vendors. It should take a cue from its actions in the directory space. The company should also consider a general purpose rules editor so all types of rules-based policies can be stored in the Registry. It should also consider basic workflow to automate IT governance operations such as rollout of services, lifecycle changes, or other operations.
Top
Positives and Concerns
Competitive Positives
• Sun announces its new Sun Service Registry, which will go into full beta next month. The product combines UDDI with an ebXML repository, and gives administrators a place to centralize management and governance of service oriented architectures (SOAs). Although others have UDDI registries (e.g., IBM, BEA, etc.), Sun has a vision of a repository, where all sorts of information about services can be stored and accessed by customers, third-party applications, and others. It also wants to link its registry to its various JES middleware elements. Sun, in effect, wants to duplicate in SOA registries what it does for directories in the identity management/security space.
• The registry combines UDDI 3.0 registry (with basic discovery) with an ebXML 3.0 repository. In addition to being able to store all kinds of metadata and “artifacts” about services (transformations, XML schema descriptions, BPEL, etc.), ebXML can be “federated” with other ebXML repositories. This means users can do queries that span registries from different vendors (as long as they are ebXML 3.0-compliant) and provide management of services in an extended network.
• Sun is building out a number of SOA governance features in its product. For example, it has an access control feature and a content-based event notification feature, as well as a feature to version services. Users can also define taxonomies and construct associations and classifications of services. Other planned features, such as cataloguing and validation, will be accommodated through a plug-in. While not all of these features are fully developed with UI and templates (the first version of the product is, after all, not even in beta), the range of features will help administrators get a rough handle on their SOAs and do some basic management. It puts them ahead, at least at this point, with competitors such as IBM and BEA.
• The firm will integrate its registry with other JES components, making the entire middleware suite more attractive. For example, there will be integration with Access Manager, so security rights can be stored or called out to form the registry. The firm’s JES Studio development tool will also allow developers to automatically register services in the registry, and a portal feature will allow portlets to be registered. Again, these will roll out over the next 12 months, but will give Sun an advantage over middleware that is not integrated into a services registry.
• The Service Registry could be a platform for all types of ISVs and could tie into ISV applications. This can include management vendors and security vendors who want to put policies and QoS-type rules in a central place. It can also include application vendors who have included out of the box Web Services and want to register them to promote good IT governance. If Sun approaches the market as a neutral party, it could help propagate its own registry and therefore its middleware, similar to how it has propagated its directory.
Competitive Concerns
• The Sun Service Registry product is not even in beta. Many of the features (e.g., user interfaces, templates, etc., for all the features) have not yet been fleshed out and will likely arrive in the next version.
• Further integration with other JES components will arrive in a piecemeal fashion over the next 12 months, giving competitors a chance to respond.
• In order for federation to work, ebXML 3.0 has to be more widely adopted. Although there is an open source ebXML available (that is the origin of the Sun product), vendors such as IBM, BEA, and Oracle have yet to actively support this idea. Some vendors are taking the stance that UDDI 3.0 has enough extensions to allow rules and artifacts to be stored there (see “SOA Software Provides Window into Services Management,” April 4, 2005). It is unclear how this is going to play out, and users may wait.
• Sun’s ISV program has not yet matured (although, this is not unusual at this stage). The company has not yet identified possible ISVs that could drive this market.
• The company is lacking a general purpose rules editor/rule engine for executing policies, and it is lacking a workflow tool to automate simple governance procedures. Smaller firms already have such features, which could be helpful in making the Service Registry a platform for storing all sorts of policies around services, including QoS.
Top
Recommended Actions
Recommended Vendor Actions
• The company should hasten to mature this product with user interfaces and templates, as it could lose its advantage as competitors enter the space.
• Sun should take a cue from its directory marketing, and offer free versions to ISVs, especially those in the management and security space. This could ultimately help drive sales of JES.
• The firm should expand features for the input of all sorts of policy rules (i.e., a rules editor/engine). It should also include a low-level workflow tool for services rollouts, versioning, and lifecycle changes. This will broaden its appeal to ISVs and IT managers.
• The firm should also consider service packs to facilitate integration with other JES elements, particular security and portal.
Recommended Competitor Actions
• Competitors in the middleware space (IBM, Microsoft, BEA, Oracle, SAP, etc.) should investigate the feasibility of a combined UDDI 3.0 and ebXML 3.0 registry. The advantage of ebXML is that it can be federated, more artifacts and policies can be stored, and it is freely available in the open source world. Other vendors should particularly watch the movements of IBM and Microsoft in this space to see if a clear trend develops.
• Competitors in the integration/SOA space should consider putting information about services in a UDDI 3.0 directory to help in IT governance of services. Beyond simple discovery and query, it should consider a central registry as a way to set policies around performance, QoS, and simple management. If these vendors don’t want to license a UDDI directory, it should make it easy to hook into or call out to UDDI 3.0.
• SOA management vendors should also consider centralizing control and administration of services in a UDDI/ebXML registry, or at least a UDDI directory. Several vendors in this space (Infravio, SOA Software, Systinet) have already gone down this route. These vendors need to add value on top of the basic spec, in the form of interface tools, rules editors, templates, links to their execution environments, etc.
Recommended End User/Customer Actions
• Users who have (or will have) a proliferation of services (over 50) should consider UDDI 3.0 for basic discovery, query, and access control to services. It is likely that the number of services will do nothing but grow.
• Users should consider UDDI and/or a UDDI/ebXML 3.0 combination as a platform for more in-depth management and governance of services, including setting complex policies around access, QoS, and storage of metadata about services and messages coming out of services.
• While UDDI 3.0 is assured as a standard, ebXML 3.0 is not. Some users may want to wait to see if the big players such as Microsoft and IBM sign off on this repository, or continue to ship only UDDI 3.0 directories.
|
