Evangelizing Mainframe
Print Email

The Mainframe and Cognitive Computing

Say the words “cognitive computing” to most people and they will think of IBM Watson. And say “IBM Watson” to most people and they will remember its success playing a TV game called “Jeopardy!” But there’s much more to IBM Watson than being able to play games, and there’s a role for mainframes in the future of cognitive computing.

Cognitive computing systems work differently from the kind of computing systems we use every day. When I’m paying an invoice through my bank (probably using a mainframe application in the background to make it happen), I want certainty in the result. I want exactly the amount that I specified to come out of my account and make its way to the account of the company that I’m paying. No more, no less. But not everything in life is quite as clearly defined as paying bills and taxes. For many things there is a degree of uncertainty. For example, if I own a supermarket, should I stock up on ice creams or warming hearty soups. It depends on the weather. What I need is a computer that can handle ambiguity and uncertainty; I need cognitive computing.

A cognitive computing system can take the available information and combine it with what are called influences, contexts, and insights. Because of the ambiguities and uncertainties and also possible conflicting information, the answers produced are always the ‘best’ possible answers rather than the ‘right’ ones. And, it will suggest how confident it is with this as an answer. This is what IBM Watson was doing to come up with its answers for “Jeopardy!”

Looking even further into the future of cognitive computing, in 2014 IBM unveiled what it called the world’s first neurosynaptic computer chip—describing it as a processor that mimics the human brain’s computing abilities and power efficiency. TrueNorth, as it was called, was designed to understand its environment, handle ambiguity, and take action in real time and in context. IBM said that the chip could be used to proactively issue tsunami alerts, do oil-spill monitoring or enforce shipping lane rules. Other possible applications include powering small search-and-rescue robots, helping visually-impaired people move around safely, and automatically distinguishing between voices in a meeting and creating accurate transcripts for each speaker.

There are two advantages with this chip:

  1. It consumes very little power.
  2. It’s a non-von-Neumann design.
Watson, like most other computers, uses von Neumann architecture, which was originally conceived by John von Neumann back in 1945. It basically has an input device, a CPU and an output device.

So, let me think, where do I have lots of data, often conflicting, that I might want to use to make sensible predictions? The answer is obviously big data. Going back to my supermarket problem from earlier, I could check customers’ shopping patterns to see whether they buy ice cream in bad weather or buy warming soup on hot days. Using my cognitive computing, I can come up with the ‘best’ things to order to have in stock in my supermarket.

Now not all big data is stored in Splunk or Hadoop or whatever. A lot of that useful information can be found on the mainframe. It might be in DB2 databases or, perhaps more likely, in IMS databases. IMS has ‘Full Function’ databases, ‘Fast Path’ databases and ‘High Availability Large Databases (HALDBs)’. They store an incredible amount of data, which can be accessed quickly and which could be fed into IBM Watson for it to use to give the best answer possible to any questions. It could use the data to predict whether customers are likely to borrow money, if my company were a bank. Or it could be used to predict shopping patterns. It could also be used to make suggestions of other products that shoppers might like to buy.

The amount of information stored on an organization’s mainframe will definitely help a cognitive computer to allow a company to make better business decisions in the future. And this will give the company a business advantage—something every company is looking for.

Trevor Eddolls is CEO at iTech-Ed Ltd, an IT consultancy. A popular speaker and blogger, he currently chairs the Virtual IMS and Virtual CICS user groups. He’s editorial director for the Arcati Mainframe Yearbook, and has been an IBM Champion every year since 2009.

Posted: 3/15/2016 12:00:19 AM by Trevor Eddolls

Print Email

Join Now!