ESBs in 2007: Taking the Open Source Bus to SOA (Part I of II)

In August 2004, I wrote that the "bus does not stop here." I explained that enterprise service buses, the 21st century form of message oriented middleware (MOM), were becoming popular underlying middle-of-the-stack software. But they were not being used very often free standing, that is, just to replace enterprise application integration (EAI) software for example. More interesting, many users were starting to build a service-oriented architecture (SOA) without an enterprise service bus (ESB). It wasn't logical but the marketplace rarely is.

The next time I looked, about 18 months ago, that pattern still held. I did find that the number of "bus routes" had grown however. By that I meant that suppliers that had originally offered ESBs only underlying their other midstack software had begun to offer ESBs a la carte as well. IBM and BEA were leading examples of that trend. Still the suppliers' preference was to sell you their 'higher order' software, enabled by the ESB, rather than just sell you the ESB. "The bus still did not stop" with just the bus, and therefore the ESB market as a separate entity was not booming. IBM and BEA telegraphed their new ESB plans for a year before they announced, which probably held ESB adoption back. (Fear, uncertainly and doubt works throughout the information technology [IT] market and is not just an anti-open-source-software (OSS) phenomenon.)

But something else was happening in the user community that I couldn't pin down back when I last took a look. So I looked again last month. The trends I saw 18 and 36 months ago seem to be continuing but I think I now see the reason for user hesitation when it comes to freestanding ESBs. I think the OSS community is the culprit.

User Waited To Avoid Lock-in, Lower Costs
Quietly, multiple OSS ESB projects started up in the 2003-2004 timeframe. They were incubating while Progress, TIBCO and others traded PR releases about which supplier built the first ESB. OSS communities and sponsoring suppliers jumped in with OSS ESBs, almost as soon as the concept took form. In my 2007 MOM/ESB research, I found almost a dozen ESB/MOM OSS projects that began in 2003 or since and that have even matured from the community stage to productization. The newcomers include

  • Bostech ChainBuilder (based on Open ESB community efforts)
  • Covalent (providing support for Apache ActiveMQ)
  • Iona Fuse project (based on code acquired with LogicBlaze and Iona's own previous OSS ESB effort, Celtix)
  • Mule project and Mulesource
  • Optaros (providing support for multiple OSS products)
  • OW2's pending project
  • Red Hat (via its 2006 JBoss acquisition)
  • Sun (via its new OSS policy and indirectly via the 2005 SeeBeyond acquisition and Project Open ESB effort within the Java Community Process)
  • wso2 (based on Apache Synapse)

Now those new style buses are starting to roll out of the car barn. If you were waiting in the bus shelter in order to base your SOA framework on a freestanding ESB here come new buses with no fare boxes. ESBs from the 'OSS line' are now in service. Reducing vendor lock-in and lowering costs are two of the more important reasons I hear about when users are making a decision for a freestanding ESB (or, less often today, a MOM). And some IT folks simply like the choice that any freestanding software offers by design.

Although built-in ESBs (e.g., the primary or only way BEA, IBM, Microsoft, Oracle and SAP sell AquaLogic, WebSphere asynchronous products, MSMQ, Integration Server and NetWeaver respectively) are convenient, they often lack the wide array of functionality of a freestanding ESB/MOM. And even when a built-in is as functional as a comparable freestanding ESB/MOM, the former are clearly designed specifically to work with the product of which they are a part. They often have APIs or types of connectors to support interoperability but interoperability is not their reason for being. If the move to an ESB-enabled SOA has been held back by the built-in nature and/or price of the traditional providers' ESBs and MOMs, this is where OSS ESBs and MOMs come in. (By traditional suppliers, I mean the big five mid-stack software suppliers mentioned above as well as Axway, Progress, Software AG and TIBCO).

OSS helps in two ways:

  • OSS lowers the initial cost (if not the total cost over a typical project's lifespan). OSS buses really do have a fare box, but it is often hidden because of the subscription/SaaS manner by which most OSS products and services are priced.
  • The OSS movement will also force down the price of the traditional providers' products and services

Of course, some users can avoid all but IT-staff costs if they forgo an OSS ESB's related service offering and provide the support themselves. That's one of the key choices that OSS provides IT staffs. But going totally "free" with OSS ESBs is not a viable option except for the largest enterprises with large IT staffs and long experience with messaging software. Forgoing formal service and support relationships does not make sense given the criticality of most applications in which an ESB would be used.

Not only are most users eventually choosing some of the new OSS ESB/MOM suppliers' services or one of their traditionally licensed "enterprise editions," but a separate services business is emerging as illustrated by companies such as Covalent and Optaros. For more detail, see the next article in this series.

The OSS ESBs and MOMs Are Out of the Incubator
Here is a summary of my findings for some of the key new-guy OSS ESB/MOM suppliers mentioned above. The illustration shows a generic version of a MOM/ESB but of course not all suppliers listed here offer all the functionality in the diagram. The more established players, including the big-five software suppliers will be discussed in the next article in this series.

Bostech ChainBuilder ESB
The Bostech ChainBuilder ESB is based on the Java-centric Open ESB community using Eclipse as the framework, Apache ServiceMix, and Apache Derby as the embedded database. Tomcat runs the Admin Console. Bostech reports over 7000 downloads with over 25 production sites fully in service already for this offering, which went GA in the spring of 2007. In its marketing, compliance with open standards is cited along with ease of development and monitoring, rapid deployment of solutions through pre-built components (especially the mapping and transformation of vertical standard data formats like EDI X12 and HL7) and support for integration of non-XML and non-Web Services applications. Bostech is of course interested in seeing ChainBuilder bundled into other popular software. Partners and products currently committed to embedding ChainBuilder ESB include CommercialWare's CW Integrator, Configure One's Concept Connect for manufacturing and its own LINC for Amazon. Bostech's attention to industry issues both in its pre-built components and choice of partners is a smart move.

The Bostech ESB, branded ChainBuilder, is available under the GPL license with professional services subscriptions and via a commercial license for partners. ChainBuilder runs on both Windows and Linux.


Iona is an ESB/MOM provider that quickly moved through all three stages of the market I described above: bundled ESB (in Artix), to freestanding ESB, to OSS ESB (the Celtix project). Artix and Celtix both have pieces of ActiveMQ in them. In March of 2007, Iona acquired LogicBlaze-another of the 2003 era OSS ESB/MOM startups-and brought all of its efforts together under the codename Fuse.

LogicBlaze hopes to be better able to deliver enterprise-level support to companies deploying its OSS technologies because of Iona's long experience supporting mission-critical applications, especially in the telecommunications industry and financial services. LogicBlaze's 25 customers are in a similar mix of industries. LogicBlaze had "incubated" out of Simula Labs founded by Winston Damarillo, also the founder of Gluecode. See a future article on OSS application and web server software. Ironically, with the acquisition, Iona becomes a more direct competitor with Cape Clear (see next article in this series), founded by Annrai O'Toole, who was also an Iona founder.

MuleSource Mule
The MuleSource OSS effort reports more than 750,000 downloads with more than 1000 sites in production, less that one year after launch. Of course Mule itself has been in the community since 2003. In its marketing, MuleSource cites the size of its community (it thinks it's the largest), what it calls adaptive integration (offering multiple types of interoperable interconnections of data, code, web sites and so forth), simplicity in development (Mule is a big enabler of Spring applications), and Mule's support for a large number of standards, protocols, and other technologies (e.g., through its community, an IBM CICS connector is available).

Mule is not bundled into other suppliers' products yet (at least not through a relationship with MuleSource). The functionality is available totally OSS or a part of standard licensed Enterprise Edition. It runs on any platform that can host a J2SE 1.4, 1.5, or 1.6 compliant Java Virtual Machine (JVM).

Red Hat JBoss ESB
The JBoss ESB effort reports 30,000 downloads to date but its primary purpose is to underlie the JBoss Enterprise SOA Platform targeted for end of CY 2007 availability. As such it will make a good research experiment on whether OSS users want their ESBs freestanding or built-in, because Red Hat offers that choice. (Remember, although it is still OSS, being locked in to an OSS suite is not that different than being locked into a suite from the big-five software suppliers.) Red Hat JBoss cites its declarative service creation with auto registration as a key feature along with a rules engine to enable content-based routing/filtering, service-level load-balancing and fail-over and what Red Hat calls business process management (BPM)-capable service orchestration and human task management. At present, Red Hat does not plan to provide a separate service plan for the ESB but it will continue to be available as an OSS project. It runs on Linux of course, Windows, HP-UX, and Solaris.

Although the ESB is new to the market, JBoss is not new to the concept. The JBoss Application Server includes JbossMQ, an integrated JMS server, and JBoss also had a Messaging 1.0 product in alpha release before the acquisition.

Sun currently makes available the Sun Java System Message Queue both freestanding and as part of the Java Systems Environment (JSE), which is its major means of offering its software since 2006 when it went totally open source. In the JSE, Message Queue 3.5 is bundled with an application server, tools and a directory server. Sun is only loosely involved with the JCP's Project Open ESB, which implements an ESB runtime using the Java Business Integration (JBI) Java Specification Request (JSR) as the foundation. This is primarily used for integration of web services to create loosely coupled enterprise class composite applications.

After multiple-year Apache incubation, the wso2 ESB was launched in the spring of 2007 coincident with the general availability of the Apache Synapse project; wso2 was a major community driver. wso2 offers other middleware products such as an application server also based on Apache. Key features of the WSO2 ESB emphasized by the company include a non-blocking http/s transport to perform virtualization and support scaling; proxy services such as transports, interfaces, message formats, quality of service, and optimization switching; an integrated registry/repository that facilitates dynamic configuration; and built-in support for external registries.

The wso2 ESB is 100-percent open source and works with J2EE, .NET, JMS, and HTTP/S-based systems, as well as Apache Axis and Axis2 endpoints.

Summary: More New Buses Will Be Added
The major question I continue to ask is whether ESB/MOM has a future as a freestanding function. I think it could happen in one or more ways:

  • Freestanding functionality will be helpful is mobile and wireless. If freestanding ESBs provide better, more cost-effective support for loosely coupled systems and Internet technologies that are the underpinning of the wireless world than they would be preferred over integration servers with broader functionality but higher overhead.
  • As with other midstack software, the future success of ESB/MOM products will involve industry specificity and support for specific roles (e.g., the increasing number of business analysts performing programmer like functions).
  • ESB/MOM Products specifically designed for small and midsized organizations also have a place.

The combination of supporting the wireless revolution, different roles, and different size companies means the number of buses will likely grow on the OSS line.

About the Author

Dennis Byron brings three decades of analyst experience to his role as ebizQ's Community Manager for Improving Business Processes. This community covers Business Process Management (BPM), Process Modeling, Process Analysis, and Business Alert Monitoring (BAM), among other topics.

As Community Manager, Byron will blog and podcast to keep the ebizQ community fully informed on the latest news and breakthroughs relevant to enterprise BPM. Byron will be responsible for bringing you breaking news on BPM daily, writing feature articles and sourcing content from other analysts, industry associations and vendors for publication on ebizQ. Finally, each week, Byron will compile the most important news and views in an e-mail newsletter for ebizQ's ever-growing BPM community.

Byron is ideally suited to the job, as he has researched and analyzed all areas of IT and information-systems use for the past 30 years. Byron looks at BPM market dynamics backed up by facts, while taking into account the perspective of the IT and business person. He is a frequent speaker and moderator on business processes, which will also be one of his roles as Community Manager.

Byron was the ERP and Middleware Analyst with the Datapro division of McGraw-Hill and IDC from 1991 to 2006. In these roles, he was the primary analyst for Business Process Management. He has conducted over 500 specific information-systems case studies. He has contributed to Application Development Trends, IT Business Edge, Research 2.0 and other publications.

Byron is also the principal of IT Investment Research, which is aimed at institutional and individual investors in IT, or anyone who enjoys peering under the covers of "the financials," where large companies and emerging IPOs like to bury their most interesting facts. His main area of interest is investment opportunities in enterprise software.

More by Dennis Byron

About ebizQ

ebizQ is the insiderís guide to next-generation business process management. We offer a growing collection of independent editorial articles on BPM trends, issues, challenges and solutions, all targeted to business and IT BPM professionals.

We cover BPM standards, governance, technology and continuous process improvement, as well as process discovery, modeling, simulation and optimization, among many other areas. We follow case management, decision management, business rules management, operational intelligence, complex event processing and other related topics. We closely track important trends such as the rise of social BPM, mobile BPM and BPM in the cloud. We also explore BPMís use in functional areas, such as supply chain and customer management, and in key verticals, such as financial services, health care, insurance and government.

ebizQ's other BPM-oriented content includes podcasts, webcasts, webinars, white papers, a variety of expert blogs, a lively online forum and much more.