CA teaches old databases new tricks
- By Stephen Swoyer
Old mainframe databases don’t fade away—they keep right on working. Take IMS, for example: Thirty years old, IMS remains a workhorse, supporting many of the same applications it powered when it was first released.
The same is true, albeit on a smaller scale, for another pair of pre-relational stalwarts, IDMS and Datacom, both of which are still being enhanced by CA. How many new tricks can you teach an old pre-relational database? Plenty, CA officials say—especially if customers are still actively banking on both platforms.
“It’s very interesting because when you really get down to it with customers, it’s not about how sexy the technology is. It’s about whether it’s meeting their needs and delivering value to the organization,” says Greg Beety, director of product management with CA’s IDMS organization. IDMS and Datacom continue to do just that, Beety insists, with the result that customers—or executive decision-makers, at any rate—continue to bank on both platforms. “We were at an executive lunch with a lot of top customers in Canada. After dessert, [an executive from one company] looks at me and says, ‘We’ve determined that this IDMS application is worth $200 million to us. It’s a $200 million application in our infrastructure. So I want to know what CA is doing about the future of IDMS and what CA is doing to invest in this so I can continue to rely on it.”
If any proof was needed, CA last week announced updated versions of both platforms. The new CA-IDMS release includes an express reorg capability for z/OS that’s designed to improve system availability by reducing scheduled downtime. The revamped IDMS also provides enhanced compliance and audit reporting capabilities, including the ability to track database activity by user IDs.
The new CA-Datacom release likewise delivers a host of availability improvements—including support for online data reorganization, online index defragmentation, uninterrupted service during scheduled IPLs, and an automated transition to an available “shadow” CA-Datacom environment if a system failure occurs. In addition, CA has introduced new Datacom features, such as parallel index queue processing and variable log block optimizations.
More than 1,000 active customers still run either IDMS or Datacom, according to Mark Combs, senior vice-president and GM of CA’s products group, including such global household names as British Telecom (BT). “They’re very active customers, very big names. They range from huge government agencies to big banks and financial institutions to major manufacturing concerns. Any company that computerized early and went to the mainframe early, in the 60s or 70s, has lots of applications written for these platforms.”
As a result, Combs argues, both platforms remain safe bets for the future. “The reality is that it’s almost never economical to go back and redo something that’s there and works. These databases all provide extremely high reliability, very good up time, and they can handle tremendous transaction volumes.”
Market watchers aren’t as sanguine about the future of IMS, IDMS, Datacom or Software AG’s Adabase, however. According to a report published last year by Gartner Inc., for example, pre-relational database market share (measured by revenue) has mostly remained flat even as market share, measured in terms of customer sites, has declined.
There’s a logical explanation for this discrepancy, concluded Gartner analyst Donald Feinberg: “This is due primarily to increased prices from the vendors, currency conversions and mainframe CPU replacement,” he writes. “In real numbers, the revenue is dropping as the number of customers and licenses decreases.”
Feinberg’s prognosis was terminal: customers should think about migrating off of IMS (or other pre-relational platforms) over the next five to ten years. “As with any aging technology, fewer people are being trained for it. Most colleges and universities are only teaching about these DBMS engines in ‘History of IT' classes. Computer science and business administration students are learning modern technologies, such as RDBMS. In addition, entrants to the job market are interested only in modern DBMS technologies. It is difficult to find trainees for the mainframe, let alone for pre-relational DBMS technologies,” he argues.
Not surprisingly, CA’s Combs disputes Feinberg’s conclusions. For starters, he says, there’s a real sense in that one of the most salient aspects of Feinberg’s critique—that the departure of expertise in older databases will make it harder and harder for companies to manage or program for these platforms—is, in fact, toothless. “Even though [the] roots [of IDMS and Datacom] are pre-relational, we invested in putting relational development into them, also, so they both support application development with conventional ANSI SQL. Insofar as the SQL is compatible—which it mostly is—you can write exactly the same query [for IDMS or Datacom] you could for any other [ANSI SQL-compliant] platform.
“I know that Gartner piece, and, unfortunately, [Feinberg] didn’t take the trouble to talk to us. But he is right about one thing: it is still necessary to have some knowledge of the older app structure in order to maintain those applications.
“The point that he misses and that he’s just completely blind to is that this isn’t new. Ever since information technology started, particularly in the IBM mainframe world, upward compatibility has been provided by the hardware vendor. There are literally applications running today that were written in languages 40 years ago. So he could have written that piece and instead of talking about databases, he could have been talking about COBOL.”
Programmers are a highly adaptable species, Combs observes, and can usually pick up new (and often very different) programming concepts and methods very rapidly. In a certain sense, he argues, that’s what code jockeys have always done. “You don’t find experienced programmers with a lot of experience out of college with anything except Java these days, so of course you’re going to have to teach people something. But just as I learned it 30 years ago, they can learn it today, too.”
Stephen Swoyer is a contributing editor for Enterprise Systems. He can be reached at firstname.lastname@example.org.