Even if the announcement that Apache Software Foundation (ASF) granted Synapse separate, top-level project billing last week didn't generate major waves of uproar of reaction and commentary, it remains a significant move to the overall state of a burgeoning open source SOA arena. Initially, after first catching wind of Synapse in early spring of last year, I was thrown off slightly by some of the divergent descriptions/understandings of the effort that were floating around the web and as is the case with incubated open source projects it was a rapidly changing code base looking to grow in a more concisely defined direction over time. Since this is normally the case with early stage open source efforts especially those that set out to tackle broad scale areas of competency like SOA, visible progress of this variety being made is encouraging for the future.
As of the current 1.1.1 release, Synapse is positioned as Paul Fremantle, Vice President of Technical Sales at WSO2 Inc. put it "a high performance, easy-to-use, Open Source Enterprise Service Bus (ESB)," an upgrade from the lightweight XML and Web service mediation framework that it was earlier in its existence when it provided three main functions: managing virtualized connections, service management and message transformation. Previously, it was useful for exchanges made through SOAP-based Web services where management of the exchanges were available through the WS-* protocols. Now support has been extended to numerous open standards such as HTTP, SOAP, FTP, SMTP, XML, XSLT, XPath, JMS, Web Services Security (WSS), Web Services Reliable Messaging (WS-RM), and more.
The above helps Synapse cover the key segments of service oriented architectures, namely:
- Execution and integration services provided through interfaces.
- Message routing, transaction, and related services/interfaces used by services.
- Service assembly foundation.
- Management, governance and control capabilities.
Plus, as an ESB the above is offered in a familiar manner that more amenable than the aforementioned mediation framework identity. It was a tad bit strange when the initial explanations of what Synapse brings to the table went, it looks like an ESB, works like one...but isn't quite an ESB. I never figured out whether that was supposed to keep the Yet-Another-ESB tag at arms length while its identity was formalized (through incubation stage) or what? I'll go out on a limb and assume that perhaps ASF noted this distinction to prevent the which one is it...Synapse or ServiceMix conversation from flaming out of control. Yet from what I understand about both, ServiceMix is built atop JBI while Synapse is designed as a simple, high performance ESB that prizes low overhead and scalability minus the JBI foundation.
Either way, now that it's clear that Synapse is in fact an ESB, the challenge for Apache is to cut through this cloud of lingering confusion and carve out space in enterprise application development tool sets. WSO2 stepped forward some eight months ago and ran its own Synapse based ESB against ServiceMix and Mule ESB in two rounds of performance testing, which provided an idea of what Synapse powered platforms can do. Perhaps this should be repeated with direct participation of members from each party (ServiceMix, Mule, Synapse) plus some of the guys from Open ESB? Providing hard evidence that a well-tuned Synapse can stack up to its open source peers tuned similarly for performance might serve as a nice coming out party/current state of affairs for the project.
Regardless, the driving force behind the Synapse value proposition remains the ASF software development process itself since it's no secret that the ASF is a leading example of leveraging open source in the creation of complex infrastructure. I'm looking ahead for the newly promoted Synapse project to capture more contribution from a wider range of enterprises, vendors and SIs. Even if given the ASF's operating model and the inherently complex nature of a category like SOA, sustained momentum won't hinge on the availability of customer-supportable products. To fill in the gaps, I expect that ISVs and SIs will handle the complex and time consuming tasks associated with integration work. The net result: an Apache Synapse that's well positioned at the forefront of what remains the early stages of SOA technology maturity.