Podcast with Gregg Willhoit: SQL, SOAP and the zIIP

In this podcast Gregg explains how the zIIP exploits data. The podcast runs for 3:12.

To listen to the podcast, please click on the following link: http://blogs.datadirect.com/media/GreggWillhoit_SQLSOAPzIIP_7.mp3

Gregg Willhoit:

I think that the zIIP is definitely coming around to exploiting data. The most ubiquitous API in the mainframe world – and the data processing world for that matter – is SQL, not SOA. So it is extremely important for a product that reports or advertises zIIP offloads to be able to just as effectively offload their SQL workloads as they can their SOA workloads.

I think DB2 does a great job of offloading SQL for DRDA type callers. What Shadow does is it offloads its SQL access to all the various non-relational database types that are found on the mainframe. And it does so as efficiently and as effectively as it does for SOA. Internal testing has shown that basically 99-100%of all the SQL related processing, the transformations, the joins, the computational intensive processing that is needed for SQL‘s non-realtional is offloaded to the zIIP. And I think this particular fact is of great of importance going forward with regard to, (1) continuing the growth and the usage of mainframe data where it sits today and in the form that it sits. As well as (2) increasing the access to said data, because it can now be done in a cost effective fashion; whereas, heretofore, it was fairly unwieldy. The data had to be moved off the host, it had to be moved somewhere else, and so on and so forth.

We have the single product which provides for both SOA and SQL. We know that GT has a partnership with a company called BOSS Software. They’re a distributor. They can’t offload that stuff to the zIIP. It’s a completely different architecture.

SQL Access to non-relational data opens up a whole new type of client to what has typically been a very cloistered environment with regard to data on the mainframe. There is no facile method and no common method for accessing VSAM, IMS, Adabas, etc. These are all different databases with different structures, different levels of integrity, different capabilities. Yet the most ubiquitous API that exists in the data processing world is SQL, and there’s no commonality across these various backends. What we’ve done is we have developed an ANSI-92 SQL engine that does access these backends using the same ANSI-92 SQLAPI.

Now what this means is any business intelligence type products that uses or generates SQL are going to backend, such as DB2, to do data analysis, data mining, the kinds of things that help businesses make the right decisions in terms of moving forward can now be done against non-relational data. And that’s where most of the data in the world is. The critical data of record, much of it is non-relational data.

So what we’ve done is we’ve opened it up to a whole new class of client, and we’ve done it in such a fashion by using the zIIP as a virtual fungible appliance with a coherent memory, we’ve done it in such a fashion that it doesn’t cost any money. All those SQL transforms and such are executing on the zIIP. And we’ve done it in such a fashion that keeps the data on the mainframe, where it belongs. So the data is current. It’s a win win situation.