In a technological era in which the critical nature of Enterprise IT is sometimes overlooked due to the hyperbole around social messaging volumes, this milestone reminds us that our clients' data, and everything it represents, is the infrastructure upon which your business is built.
A History of Leadership
April 23, 1987. IBM Santa Teresa Lab, San Jose, California. A single IMS system is benchmarked at 1012 TPS. The project, code named ONEKAY, was known only to the IMS Performance team at the lab and a few engineers across IBM who made improvements to their products so this mainframe milestone could be achieved. Why a secret? It was not supposed to happen. IMS was a 20-year old product intending to reclaim its greatness at a time when the industry was becoming enamored with enterprise relational database systems (RDBMS), this despite the fact that the benchmark for an SQL database in 1987 topped out at 208 TPS, roughly one-fifth of the IMS benchmark.
“ With the success of DB2, many industry watchers were predicting the demise of IBM's aging IMS product. To demonstrate that IMS is far from dead, IBM recently announced benchmark results that showed the latest IMS release executing over a thousand transactions per second, confirming the success of the work being done to enhance IMS for high volume processing... The prominence given to these results by IBM demonstrates their seriousness about repositioning IMS as a DB/DC system for handling the large terminal networks and transaction volumes that will be required over the next five to ten years. ”
- InfoDB - The Journal for Large Systems Database Users. Colin White, Summer 1987
Now, twenty-five years later, IMS's transaction processing competitors from that era are faded from memory, no single-image RDBMS has ever achieved the transaction rates of IMS Fast Path on comparable configurations and quality of service, IMS usage is at its highest level ever, and IMS is reaching for new goals.
Perspective, Hindsight, and What's Next for IMS
Perspective is essential when considering benchmarks. Ten years after IMS contributed to the successful Apollo moon shot, the IMS Fast Path 1978 benchmark drove approximately 88 TPS on a Model 168P, and 100 TPS on a 3033UP class processor. The 1987 benchmark was on a 3090-400 class machine with 128MB of central storage, 12 3880-3 control units, and 33 3380 direct access storage devices that required, in total, at least 600 square feet of floor space.
By comparison, a moderately sized IBM zEnterprise EC12 and a single IBM Storage Solution DS8870 can drive 100,000 TPS in less than one-tenth the amount of floor space (imagine the size of your standard cubicle), a fraction of the power, still on a single IMS system, and require no more administrators to accomplish 100 times more work.
The IBM zEnterprise is designed to support many concurrent workloads. Today's largest single zEnterprise EC12 processor could drive an additional 10 times as much work. When IMS first achieved 1000 TPS in 1987, no client required those volumes, but now these rates are common. Factor in the demands from our increasingly Smarter Planet, including mobile devices and analytics that will soon drive rates of 10s of thousands of TPS, with 100% availability, reliability, and security. How soon will you need 100,000 TPS? It might be sooner than you think.
Don't count IMS out just yet. We can't wait to see what happens next.
IMS 13 Migration Workshop
Coming soon to a city near you!Register today
IMS Technical Symposium
Join us in March in Germany!Register now
IMS 13 and IMS Enterprise Suite 3.1
Webcast from IMS Development LabRegister now
IMS 13: Record-setting performance and powerful data distribution
IBM’s IMS — Myths, Realities, and Opportunities