Search Rocket site

Lowering Mainframe Costs

Reduce Mainframe Total Cost of Ownership

Patented Technology Leveraging zIIP Specialty Engines

Over the years, IBM has been quietly introducing several innovations in the form of “specialty engines” to its System z architecture. These specialty engines create a powerful symbiosis with specialized software that can deliver dramatic savings in mainframe capacity usage and costs. Specialty engines are expanding the opportunity for mainframes to participate in new areas, such as Big Data, analytics, data virtualization, mobile and Cloud initiatives.

IBM characterizes specialty engines as processors that help users expand the use of the mainframe for new workloads, while increasing performance and lowering total cost of ownership (TCO). Rocket Data products and solutions have been built from the ground up to use the System z Integrated Information Processor (zIIP) specialty engine. Our solution enables data to remain on the mainframe, integrated in-place, leveraging the zIIP engine to handle all of the processing. Gone is the rationale for moving mainframe data off-host to avoid high processing cost. This approach seamlessly incorporates mainframe functionality and data into other business applications, faster, with less complexity and at the same time reducing TCO.

Rocket has a proven, risk-free method for using zIIP specialty engines. Rocket does not cause IBM or other third party code to become zIIP-enabled.

Rocket Data products can divert up to 99% of the workload that would occur on the mainframe general purpose processor (GPP) associated with data integration and application modernization to the zIIP engine mainframe.

For Rocket Data Virtualization Server (DVS), here is more about what it can do:

  • Seamlessly diverts processing-intensive large data set queries involving mainframe data sources including Adabas, Db2 for z/OS, IMS DB, and VSAM from the mainframe's GPP to the zIIP specialty engine.
  • Extends TCO benefits of zIIP to so called “ineligible” zIIP environments – like VSAM, Adabas, IMS or CICS and Natural applications, all which must run on standard GGP processors. But with Rocket DVS integration or virtualization with these environments runs in our address space, and due to our patented technology, is eligible to run on the zIIP.
  • Significantly increases the Transaction per Second (TPS) rate – often by as much as 175% by leveraging the zIIP specialty engine. Rocket helps organizations better manage mainframe capacity usage.
  • Does not cause IBM or other third party code to become zIIP-enabled

Don’t Be Fooled By Imitations

The use of specialty engines to lower mainframe TCO continues to expand. Some middleware vendors leverage specialty engines to reassign some portion of XML processing via IBM’s z/OS XML System Services (z/OS XML)—a system-level XML parser with superb performance characteristics that’s integrated with the base z/OS operating system. Rocket DVS utilizes this capability as well, but has taken a more holistic approach. Through the use of z/OS XML other mainframe middleware vendors are capable of exploiting specialty engines, but their approach is limited in its capacity to do so and omits a considerable portion of the processing involved in XML integration. In the end z/OS XML parses inbound XML into a tokenized format. However, this format is unusable by the back-end; it must undergo further conversion to whatever format is acceptable there—potentially adding intensive processing that’s left to the GPP and is subject to software usage fees. In addition, the outbound XML (substantially larger than the inbound) also can’t be reassigned to the zIIP.

The manner in which Rocket DVS is architected is fundamentally different. With its zIIP-enabled data virtualization architecture, Rocket DVS accomplishes up to 99% of the XML processing on the zIIP using hybrid threading models that dynamically switch from SRB mode to TCB mode for optimized performance and zIIP access. Unlike competing products that can’t exploit the zIIP, or that do so under constrained circumstances, this hybrid model of integration stream can enable almost all necessary processing— such as XML un-marshaling and marshaling, security processing, and development-related processing—to be migrated from the GPP to the zIIP, where it’s processed faster and at a lower cost.

The optimal design approach for utilizing mainframe specialty engines builds exploitation capabilities into the foundational architecture of the data integration software. When the underlying codebase is designed to sense and react to the presence of a specialty engine, the capacity offload benefits are holistic across all data integration scenarios – data virtualization, web services and SOA, web-enablement – ultimately translating into lower mainframe costs.