Content networking appliance maker Solace Systems pressed the case for its hardware approach to middleware with a recent high-speed message caching update for its Unified Messaging Platform. Think of it as an ESB accelerator!
Members of the Solace Systems brain trust hail from such vaunted technology power houses as Tibco, Newbridge and Northern Telecom. What they have done is created a machine specifically tuned to route messages. At the heart of the solution is a network processor chip with high I/O bandwidth and associated field programmable gate arrays capable of hard core data crunching. The company claims a pure hardware messaging path means very low (and consistent) latency, and throughput performance 10- to 20- times greater than software-centric messaging servers.
To program its platform, Solace has fashioned a unified API that supports various types of middleware methods. The Solace API now supports interaction with the high-powered cache, controlling synchronization of cache requests, supporting multiple instances for scalability, failover and load balancing.
There are lessons to be learned for SOA and middleware in the evolution of network computing, suggests Hans Jespersen, principal system engineer, Solace. Networking started with complex designs, but eventually, network switches served as reliable building blocks for network architects.
Solace’s single API is said to work across middleware types that include high-fan-out, guaranteed MQ, as well content routing and transformation types.
“You shouldn’t have to engineer it if it can be in a box,” said Jespersen. “The best practices are inside the box.”
Why haven’t various middleware types been unified before? “They were done in software and there was contention for resources,” said Jepersen. He maintains you obviate those issues when you rely on hardware.
Despite the new caching capability, the company will continue to support third-party caching technology, according to Solace Chief Architect Shawn McAllister. The new high-speed cache is said to support requests for individual topics, or groups of topics, and can resolve consistency issues when new data is entering a cache as stored data is being requested.