Joint Battlespace Infosphere (JBI) Content Based Routing

摘要

The purpose of this effort was to integrate a hardware based eXtensible Markup Language (XML) processor into the JBI architecture. Integration was to be invisible to the user and work seamlessly with the current client software. Changes necessary to support plug-in hardware solutions should not require client code changes but rely solely on JBI core framework changes. At low loading levels, the two implementations should behave similarly. The hardware-assisted JBI implementation should accelerate JBI operations under heavy loading of publication/subscribe requests with complicated XPATH parsing requirements. It was deemed that XPATH processing and XML parsing were the two areas most likely to suffer under extreme loads and promising areas to leverage commercial technology. A software-based solution must ingest the whole JBI object before beginning processing. Object sizes and number of concurrent objects processed will directly affect system responsiveness. A hardware-based solution can process objects at line speed (e.g. 1Gb/s) as packets flow through the network. Large/complex objects will potentially slow processing down, but in general, as bits flow on the wire the internal workings of the XML router is making decisions and taking action on the data. Lastly, the hardware-based solution is scalable and can provide more processing power as required.