The mainframe isn’t dead after all
August 21, 2008 
By Jeff Gould, CEO & Director of Research, Peerstone Research
Arnold Schwarzenegger and California’s Controller John Chiang are still jousting over the status of that supposedly dead computer language, Cobol. As I mentioned a few weeks ago, Schwarzenegger has ordered Chiang to slash salaries of California state workers in order to conserve cash until the bitterly divided State Assembly gets around to approving a new budget. But Chiang is saying he can’t do it because the state’s payroll program is written in Cobol and would – according to Chiang – take months to modify.
My guess is that Chiang doesn’t have the faintest idea whether his excuse is actually true. He’s just using it because it’s convenient and because he thinks no one will step forward to contradict him – the state’s long-suffering IT guys and gals are used to being blamed and bullied by the pols. But last week I had the occasion to visit SHARE, the premier mainframe conference, which was held in San Jose just down the road from where I live. Based on what I saw, there is one thing I can tell you for sure, and that is that Cobol is not dead. And neither is the mainframe.
When I mentioned to one of my friends that I had been to SHAREa few 60-somethings strolling around the halls, the under 40 generation was also well represented. What struck me the most was not the advanced age of the people but the relative youth of a lot of the software being discussed. In the past few years a whole new class of mainframe applications has emerged which bring non-mainframe environments such as Java and Linux to Big Iron. WebSphere on z, Linux on z – these are much more than mere checkbox options in some bait-and-switch IBM marketing scheme. A substantial chunk of the 10,000 or so System z mainframes in the world are actually running these environments. And they are doing it for big-time production apps too, not just development.
At SHARE the sessions devoted to Java, Linux and SOA [service oriented architecture] were jammed with mainframers of all ages. IBM’s brain trust was out in force at these events to tout the new technologies. Big Blue has always been known for keeping on staff a select group of hardware and software experts whose role is explain and evangelize the products they implement. Although they really serve more as marketing mavens than as actual developers, the “Big Blue intellectuals” present at SHARE were as articulate as ever in expounding the benefits of bringing Java and Linux-based “new workloads” to the mainframe. And it seems that the customers are getting the message, or at least they are getting their checkbooks out. According to unofficial estimates, Linux and Java account for a third to a half of all the mainframe MIPS that IBM sells these days.
But what about Cobol? Surely IBM isn’t touting that as a new technology? Well, no, not exactly. But believe it or not they are promoting it as a viable platform for SOA. One particularly noteworthy presentation at SHARE delved into the intricacies of generating XML messages in Cobol. It can be done, and apparently more than one mainframe customer is doing it.
So there, Steve Chiang. Your Cobol-ate-my-homework excuse is starting to sound a little feeble.
IBM Mainframe | 1 Reference 
Reader Comments (13)
Jeff,
As you point out, John Chiang has no idea how long it would take to modify the COBOL programs in question. I worked at the State Controllers Office during the time the then-new state Personnel Information Management System (PIMS) was developed and written, and wrote many of the programs in the system, and later was involved in performance tuning and security analysis of the system. I know for a fact it would take just a few minutes to implement the requested changes, since the system is table-driven for the salaries. Of course, this isn't the point John Chiang is trying to make. He is involved in a political throw-your-weight around contest with the governor.
Not only is COBOL not dead, but the mainframe systems that underpin the world's financial systems are primarily written in COBOL and essentially depend on it.
My daughter Kristine Neely is the project manager at SHARE of the zNextGen project, a project devoted to introducing young profesionals to the world of the mainframe.
Tom Harper
IMS Utilities Development Team
Neon Enterprise Software, Inc.
Sugar Land, TX
Jeff, Tom -
Well said. I concur. I often find that technical problems are a lot easier to solve than political ones. Especially self-induced ones.
Andy Barreras
Well, the thing IBM should do, if it really wants to see more IT professionals looking seriously at the mainframe, is start giving a lot more prominence to the Hercules mainframe emulator, and releasing OS390 and VM/ESA etc., disk images for it. (It was once available - from what little I know - but IBM then froze up on it, and shot off their own foot, since the approved mainframe emulator - not Hercules - was sold in a package that cost a lot and was more hassle than it was worth.)
It is nice to have confirmation that the payroll system is table-driven from Mr. Harper; I could not imagine an reasonable way to do it. But let's ask Mr. Chiang the next logical question. If the Governor had given a pay raise, then would he have found a way to implement it?
How would they handle deductions, where some employees have more deducted from their check than $6.55 would cover? How would they handle taxes on the $6.55? When the difference between minimum wage and the real wage was paid back, how would they handle the taxes and deductions on that?
All unanswered questions. How do you program for that?
Hoag
You blew your credibility right there at the end. "...generating XML messages in Cobol. It can be done, and apparently more than one mainframe customer is doing it". As every programming beginner knows, XML is just structured text; and text string manipulation is something COBOL can certainly hack. Indeed, there is no useful programming language that can't be used to create XML documents; if it can print "Hello World", it can also print (or save) XML.
Tying two of these comments together, I wonder why Hercules as a reliable mainframe emulator isn't featured as part of the zNextGen project? I would have thought it would be in IBM's interest to have a low or no cost way to experience zOS. Why not license zOS for Hercules?
Mr. Welsh, the "XML GENERATE" statement will generate XML from an existing COBOL record structure. That bypasses a lot of the tedium that a "Hello world" approach to the problem would entail. Mr. Gould's credibility is quite intact.
The point about COBOL processing XML was completely valid. Real XML parsing is not just simple string manipulation. As implemented by SAX and DOM, XML parsing is quite sophisticated with event handlers, support for various language encodings, validation against a schema, and more.
COBOL does *real* XML parsing and, beginning with V4.1 (4Q 07), it uses z/OS XML as the integrated parser, which offers ultra high performance parsing and zIIP redirection.
I talk to a lot of customers who can't understand why the other systems can't do all the things they've been doing with their mainframes - the reliability, scalability, performance, security, support, etc. just aren't the same on anything else.
Hoag,
Your questions are all valid ones, but these situations have arisen many times before, so the system was designed from the beginning to handle just such cases. While I was a state employee, this code was put to use a number of times. Retroactive pay raises are a basic requirement of any substantial payroll system.
WillDavid,
IBM has a program named the IBM Education Initiative (feel free to Google it). As part of this program, access to the latest IBM mainframes is provided to participating universities free-of-charge.
zNextGen details this in their website which is available at SHARE.ORG.
Tom Harper
IMS Utilities Development Team
Neon Enterprise Software, Inc.
Sugar Land, TX
Jeff, your post led me on a trip down memory lane, working at IBM under VM/CMS:
http://tahitiviews.blogspot.com/2008/08/im-not-dead-yet.html
John
I work on UNIX and ZOS platforms. In a large organization I found ZOS mainframe provides a homogeneous platform, whereas UNIX becomes dispersed over time. Different flavors of UNIX are implemented by individual development efforts thinking they have a better mouse trap. For relatively small organizations with limited number of server racks, UNIX is the answer but for large monolithic organizations mainframe is the economic choice.
However, for certain tasks, UNIX servers can be strategically mixed with ZOS for distributed front end processing. But once UNIX is brought into the mix, some form of middleware comes with it, not to mention another programming language like JAVA. This adds many pieces to a complex puzzle.