BEA Dresses Tuxedo Up For SOA
Type: Competitive Intelligence Report
Analyst: S. Willett
Report Date: July 19, 2005
Module: Application Infrastructure
ID: CIR14165 |
Current Perspective: Positive/Neutral
Vendor Importance: High
Market Impact: Moderate |
|
Summary
Event Summary
July 18, 2005 -- BEA Systems, Inc. announced the general availability of BEA Tuxedo 9.0, the latest version of the company's software product for high transaction processing. Applications running on BEA Tuxedo form the backbone of some of the world's leading companies running some of the largest, mission-critical transaction processing systems -- including wire transfers, Automated Teller Machines (ATMs), and telecommunications billing systems. Leveraging open standards, BEA Tuxedo 9.0 is designed to help customers build new and repurpose legacy applications and extend them to a service-oriented architecture (SOA). SOA is an approach to software application development designed to transform business applications into "services" -- specific pieces of functionality -- which can be combined and reused, helping IT organizations more closely align to business processes.
Analytical Summary
. Current Perspective: Slightly positive on BEA's new version of Tuxedo, as the firm is moving slowly to bring the legacy middleware into the world of SOAs, and its WebLogic environment, although it needs to do more to prevent defections to new middleware from application players.
. Vendor Importance: High to BEA as Tuxedo still represents a significant source of revenue and the firm needed to slowly bring these customers forward into the world of SOAs and Java application servers.
. Market Impact: Moderate on the market as there is little new technology here, but customers, especially application players with middleware stacks, will increasingly form strategies to migrate customers from Tuxedo to their own Java-based transaction processing layers.
Target Markets
B2B Communities, End Users, Global 2000, Resellers/Channels, Systems Integrators, Third Party Implementers, Web Portals
Top
Perspective
Current Perspective: Positive/Neutral
We are taking a slightly positive stance on BEA's release of a new version of Tuxedo, its legacy transaction processing (TP) monitor. The company is introducing new services enablement technology at an appropriate rate, given the nature of most TP applications. It is also introducing more links into WebLogic and Aqualogic, in a way that will attract some upselling without being so heavy-handed as to alienate these important customers.
While many industry observers predicted that Java technology (including Java Transaction Services) and other technologies would eclipse proprietary transaction processing monitors such as Tuxedo, the technology continues to have a long half life. The reason is the same reason that CICS, Non-Stop, and other legacy technologies continue to endure: they work! Tuxedo is running some pretty impressive operations including processing financial transaction, airline tickets, and other systems. While openness is always a concern, some of these are fairly discrete operations where the chief concern is 24/7 quality of service. Thus the inclusion of these applications into an SOA is not an urgent priority. Nevertheless, some benefit can be made of streaming information into and out of Tuxedo type applications through services, as long as the integrity of the application is not disturbed.
Thus BEA is taking the right tack in updating Tuxedo. The company is upgrading its Tuxedo "control" for Workshop, which will include more support for constructing services. There is an FML to XML (FML is a Tuxedo format) transformation tool, and a meta-data repository that can contain information about WSDL to Tuxedo transformations, service invocations, or other processes. There will also be a native SOAP stack directly in Tuxedo, and support for the "Aqualogic" service bus transport. The control will be available in the winter of 2005. The new Tuxedo "kernel" has also just shipped. It supports JOLT and SNMP for management as well as a number of bread and butter features for performance and management. There is enhanced support for security, including Kerberos authentication, a plug in for PKI. Its long standing mainframe adapter has also been updated with support for new versions of CICS and IMS. This keeps the product current, and meets requirements to run in heterogeneous environments. The company is also enhancing its Tuxedo Connector for WLI (WebLogic Integration), with better administration and performance.
While this will likely satisfy Tuxedo users, this is mature technology, and will mainly attract revenue from existing customers. Also, unfortunately, the new Tuxedo enhancements are delivered in a myriad of different releases, adapters, and service packs. While the core kernel has been updated, the connectors into WLI, mainframe, and Aqualogic are arriving at different times, as is the new WebLogic control. It would have been more coherent to deliver this new control in the context of version 9.0. and synchronize all these releases. While the cautious approach is appropriate for SOAs, BEA should provide more guidance as to when and how to put TP type of applications into an SOA without affecting QoS. This should be done through the consulting/professional services organization. Also, the company doesn't have many points that counter the threat that Oracle and SAP will surely bring to bear to convert Tuxedo users to their middleware platforms. BEA should preserve these users with greater support for Oracle and SAP specific APIs and middleware elements through adapters and other support. At the same time, however, the company should position its QoS features against SAP and Oracle middleware and publish performance metrics.
BEA's moves will not cause a great rush back to "proprietary" TP monitors. However, this will cause vendors such as IBM and HP to continue to enable services to interact with their TP applications, in a safe manner. Users of such applications will likely continue to slowly accommodate these to new SOAs; however, it is likely that they will not go too far, too fast.
Top
Positives and Concerns
Competitive Positives
. BEA releases a new version of Tuxedo, its legacy transaction processing (TP) monitor. The new version includes more support for Web Services, and more tie-ins to its Workshop environment, and by association the rest of the WebLogic suite. With 1600 direct customers, Tuxedo is still a significant revenue stream for BEA (number two or three depending on the quarter), and BEA is taking great care in slowly introducing new technologies that don't upset this conservative user base, but also make them feel that there is a roadmap to move into SOAs. The firm is essentially fighting a rearguard battle against Java-based middleware from rivals, and is striking a good balance between supporting new standards, while maintaining support for extremely important legacy applications from players such as FedEx, SWIFT, United Airlines, and AT&T/SBC.
. The company is upgrading its Tuxedo "control" for Workshop, which will include more support for constructing services. There is an FML to XML (FML is a Tuxedo format) transformation tool, and a meta-data repository that can contain information about WSDL to Tuxedo transformations, service invocations, or other processes. There will also be a native SOAP stack, and support for the "Aqualogic" service bus transport. This control will help users expose some processes as services, or use Tuxedo as a kind of transport between services if that is desired. While core TP processes will likely remain discrete, point-to-point type of operations, this allows users to expose some types of information as services (or conversely to bring some info in from external services) in a controlled manner. This will be available in winter of 2005.
. BEA has also just shipped a new version (version 9) that runs on a new kernel that supports JOLT and SNMP for management. The new version includes a number of bread and butter features for performance and management. This includes better control of time outs for service levels, domain connection policies, and a better framework for tracing events. There is enhanced support for security, including Kerberos authentication, a plug in for PKI. Its long standing mainframe adapter has also been updated with support for new versions of CICS and IMS. There is support for new versions of Oracle RAC. This keeps the product current, and meets requirements to run in heterogeneous environments.
. The company is also enhancing its Tuxedo Connector for WLI (WebLogic Integration), with better administration and performance. This will allow processes and data from a Tuxedo environment to participate in larger composite application that are built on SOAs. Through the Workshop control, users can already more easily build Web Interfaces or associated Web Applications or even J2EE applications for Tuxedo processes. The company estimates that 75% of Tuxedo applications have some sort of Web interface. In the long run, this helps the firm upsell its WebLogic technology to Tuxedo users.
. While the Tuxedo control seems to be the most convenient way to build services, support for a native SOAP stack, as well as transformation, will make it more than feasible to use third-party products to build and maintain services. This is a realistic move since Tuxedo is running in all sorts of environments and BEA needs to support these customers and prevent migration to rival transaction technology.
Competitive Concerns
. The new Tuxedo enhancements are delivered in a myriad of different releases, adapters, and service packs. While the core kernel has been updated, the connectors into WLI, mainframe, and Aqualogic are arriving at different times, as is the new WebLogic control. It would have been more coherent to deliver this new control in the context of version 9.0. and synchronize all these releases.
. Aside from the support for building services, the changes in 9.0 are incremental in nature. TP is a mature technology, and BEA will mainly gain revenue from existing customers, even with the new features. Java-based technologies will continue to cut into Tuxedo and in the future it is possible that some WS-* technologies could provide an alternative to lower end type of transaction applications.
. While it is clear that Tuxedo users will want to tread carefully in terms of making TP applications available as services, BEA (at this point) is not providing a lot of guidance as to when and how these legacy, mission critical applications will participate in an SOA and what controls will have to be put into place so QoS is not affected.
. BEA is not really responding to the threat that Oracle and SAP will bring to bear to migrate customers to their own middleware, away from Tuxedo. For example, it is more than likely that Oracle will attempt to lure the significant PeopleSoft Tuxedo base (over 5K customers) away from BEA.
. Support for a native SOAP stack and associated standards is a double edged sword. While it will help keep customers, it also makes it easier to bring third-party tools or even application servers in to enhance or extend Tuxedo applications either through services, or traditional Java/Web programming.
Top
Recommended Actions
Recommended Vendor Actions
. BEA is right to move slowly in supporting services into the core Tuxedo product; however, it should attempt to synchronize the release of major versions and controls and adapters. It should also improve user interfaces for transformations and service configurations (outside of Workshop interface).
. BEA should provide more guidance for Tuxedo users who want to migrate into an SOA in a secure way that preserves QoS. This should include best practices, methodologies, and samples. This will likely come from the firm's professional services department.
. BEA should counter the threat that will be posed by SAP and Oracle. Ironically, this will likely involve greater support for Oracle and SAP specific APIs and middleware elements through adapters and other support. At the same time, however, the company should position its QoS features against SAP and Oracle middleware and publish performance metrics.
. The company should continue to support key vertical markets (telco, financial, etc.) through specific APIs, format, and standards support. It should fold this either into the core product, or into vertical variations of Tuxedo.
Recommended Competitor Actions
. Competitors in the application server and general middleware space should continue to try to lure Tuxedo users away into Java-based technologies, or other approaches based on messaging. They should provide a reasonable migration strategy that ensures no disruption as users upgrade.
. While competitors should advertise support for SOAs as an option to TP users, they should be careful to keep in mind concerns about QoS and security in these mission critical applications. Support for services will be secondary to keeping these applications up and running 24/7 at optimal performance.
. Competitors should add transactional features to SOAs through relevant WS-* standards, but should be realistic as to the timeframe that applications built on this will be feasible.
. Competitors in the TP space with "proprietary technology" (HP, IBM Encina/CICS, etc.) should prepare their own interfaces into Web Services and service oriented architectures, with appropriate governance to ensure no disruption in performance.
Recommended End User / Customer Actions
. Current Tuxedo users should welcome the new release as it contains a number of incremental features in terms of manageability, performance, and platform support.
. Users should realize that many of the features for Tuxedo are actually contained in adapters and connectors (Tuxedo Workshop control, Tuxedo connector for WLI, etc.) that are being shipped out over the course of the next six months.
. Users of Tuxedo and TP systems should expose information into services (or import information through services) with a number of caveats. For one, Web Services are not, in and of themselves, transactional in nature and there is no QoS. Also, undue demands on information in a TP system could affect overall performance and this must be taken into consideration. There are also security concerns that need to be considered with a Web Services interface.
|
