Another view: ESBs are so early 2000s

A couple of days back, I speculated on the future of enterprise service buses, opining that , yes, they will proliferate everywhere, but increasingly melt into the SOA background, or meld into more expansive SOA suites.

In a recent post over at Webservices.Org, Rogue Wave Software's Patrick Leonard also mused on the future of ESBs, predicted that its role will be "redefined" as SOA deployments grow, and that more robust alternatives are emerging:

"An ESB can be a great way to deliver messages reliably from point A to B (among other things), but it isn’t best suited to serve as the cornerstone of an SOA implementation – there are enterprise-ready alternatives that are designed specifically for SOA that enable greater realization of the true goal of implementing an SOA. The Service Component Architecture (SCA) and Microsoft’s Windows Communication Foundation (WCF) are both good examples of reference architecture for an SOA. There is now product available from multiple vendors, with both commercial and open source options available for SCA."

ESBs served an important role in the early development of SOA in a world full of locked-up, proprietary systems, Leonard observed. "An ESB can provide important enterprise requirements including guaranteed message delivery and integration to back-end applications. Every technology since the dawn of enterprise computing has had to meet these requirements before being deployed in mission-critical applications." However, he added, ESBs were designed for enterprise application integration (EAI) implementations, which were meant to enable messaging between systems within a traditional architecture.

SOA begins to upend those architectures into something entirely different. Time to move on beyond the ESB, Leonard said.