Josef here is the article if you can access it:
regards,
risky
March 23, 1998, TechWeb News
--------------------------------------------------------------------------------
The Mainframe's Black Hole -- The date- conversion problem is bringing work on many other important systems to a near standstill
By Martin J. Garvey
For some mainframe users, the year 2000 crisis is turning out to be a lot more than an annoying, expensive problem. It's also emerging as a developmental black hole, sucking up money and talent and leaving other, equally important mainframe work unfinished-even unstarted. "Year 2000 isn't as strategic as some other projects," says Chuck Foley, VP of server marketing for Amdahl Corp., "but it's the one that turns into a pumpkin if you don't fix it."
John Young, enterprise system analyst for the Clipper Group, an IT advisory firm in Wellesley, Mass., goes further:"Nothing will happen between now and 1999 that doesn't involve securing systems for the year 2000," he says. "Funds will be redirected from hardware updates and new software for the next two years to support year 2000 projects. Anyone who thinks otherwise is on a mistaken curve."
If Young is right, the impact could reach beyond the IS shop and hurt companies' earnings. That's because year 2000 work is essentially maintenance. While it serves the important role of keeping the big iron running, year 2000 work won't boost productivity, it won't create new business opportunities, it won't increase market share, and it won't find new business.
To be sure, not all mainframe systems and staff are focused solely on year 2000 conversions. For example, the Long Island Lighting Co. is too busy merging with Keyspan Energy-formerly Brooklyn Union Gas-and converting its customer-information system to dedicate all resources to the date-field fix. "Because of those projects, year 2000 can't dominate, but as it is, 10% of one mainframe is all that's being used for year 2000 testing," says Jim Langton, systems integration manager for Lilco, in Hicksville, N.Y. "The other projects consumed most of our manpower, so we were lean for year 2000."
Among the biggest year 2000 tasks facing mainframe shops are converting old applications, then testing those fixes. Simply testing Cobol applications requires extra resources and better-staffed second and third shifts in the data centers. Companies running 24 hours a day, seven days a week wonder when any integration testing can take place. When internal projects are complete, attention will turn to partners and distributors.
Merrill Lynch & Co., the investment firm in New York, now dedicates about one-quarter of its total mainframe systems capacity to year 2000 testing and development. Less critical projects, such as screen-format changes and decision-support systems, are being delayed. Mike Glodo, VP of enterprise central computing at Merrill Lynch, says that percentage will grow as the year 2000 draws closer. "It's one of the nasty projects with a hard deadline," he says.
That's probably a typical measure. Industry analysts estimate that anywhere from 10% to 25% of all mainframe resources are now dedicated to the year 2000 problem. Some companies haven't even begun testing, so those figures can only rise. Moreover, companies far along with the date-field conversion are bracing for the extra work brought on by the painstaking task of getting distributors and partners prepared for 2000. "Having our own systems compliant isn't enough," says Glodo.
The black hole is also evident to the systems vendor with the largest number of noncompliant systems out there:IBM. While the company's newest mainframes are year 2000-compliant, plenty of customers still run their businesses on IBM and compatible mainframes that are more than 20 years old-machines that are anything but ready for 2000. "The draining of mainframe resources is happening," says Colleen Arnold, IBM's general manager for year 2000 global services. "Customers must prioritize where they put their resources and money, then determine which projects are continued or kicked off. By that exercise, some projects will be stalled."
However, Arnold says she hasn't seen any customers cancel important projects because of year 2000 work. In fact, she views much of the millennium diversion as a way to clean up nonstrategic work.
IBM isn't standing by waiting for the big business shutdown in 2000. Arnold points to a company Web site (www.ibm.com/year2000) that lets customers enter their mainframe model numbers and match them up to compliant systems. To be sure, the site is a sales tool, but it also offers a free database that customers can use to find out how compliant their current systems are-and to get the answer within 24 hours. IBM says the site gets as many as 215,000 hits a week, a number it expects will grow.
'Tough Couple Of Years'
The view is less sanguine at IBM competitor Amdahl, in Sunnyvale, Calif. "Customers have to be smart in laying out the workload, so it will be a tough couple of years for the application vendors," says Foley of Amdahl. "There's no time to deploy the enterprise applications this year, because customers are on the red line with year 2000. And because companies are so devoted to it, there will be no refueling in '99."
One big issue, given the current people shortage, is how to allocate staff. At Prudential Insurance Co. of America, the business groups assigned their best project managers to year 2000 work. Prudential spared the mainframe staff by outsourcing as much as 70% of the early year 2000 work. But the final business testing is being done in-house. "Companies must juggle people in for the project and then juggle them back out as they achieve success," says Irene Dec, the Newark, N.J., company's VP of corporate IT and companywide year 2000 program manager. "Year 2000 is the IS project-management work of the century."
Another concern is allocating mainframe Mips. Prudential has accommodated the additional burden by bringing in an extra CPU. Installed late last year, the machine will be used for year 2000 testing through the end of 1999. Beyond that, the company hopes the machine will ultimately support new business.
Year 2000 work also eats up a lot of mainframe Mips, so Prudential has moved year 2000 testing to the slowest times of the day, from 5 p.m. to midnight. "We're predominantly in testing now, so we hit CPUs when usage is not as high," says Dec. She says year 2000 testing draws three times the mainframe resources than other projects, in part because every solution has to be tested for at least three possibilities:before the year 2000; the minutes before and after midnight of Dec. 31, 1999; and the first workday of 2000.
Tests Are Taxing
For Prudential and many other mainframe users, this year and next will be dedicated to testing the year 2000 fixes made in the previous two years. This work could be even more taxing on mainframe resources, since many companies-even those willing to outsource the date-field conversion work-want to do the testing internally. Amdahl's Foley believes most companies will be unable to devote multiple mainframes to year 2000 testing. "Multiple dedicated mainframes sitting there are like birth certificates with expiration dates," he says. "Customers can squeeze tests in overnight, but people don't have the capabilities or the budgets for integrated tests."
Year 2000 work is also cutting into the productivity of IT spending. The average IT dollar spent last year by companies generated $38 in corporate revenue, down from $43 in 1996, according to a Meta Group survey of more than 600 companies. Year 2000 isn't the only reason-Meta Group also points to higher salaries due to the labor shortage-but it's among the main reasons. "This is the first time we're seeing inflation of the IT dollar," says Howard Rubin, a research fellow at Meta and chairman of the computer science department at New York's Hunter College. Rubin expects the trend to continue.
Indeed, salaries for year 2000 workers could continue rising. "It will be difficult to find people," predicts David Condra, president of Dalcon Business Systems Inc., a value-added reseller in Nashville, Tenn. Dalcon has year 2000 work scheduled with existing customers, but Condra still expects the issue to tie up resources and cost more money than expected. "Companies are getting anxious now," he says. "That makes people more difficult to find, and more expensive once you find them."
To be sure, there's some debate on the matter. "As much as year 2000 purists want everyone to put brakes on other projects, it can't happen, because business doesn't stand still," says Bob Cohen, VP of the Information Technology Association of America, a trade group in Arlington, Va. The ITAA surveyed a mix of systems vendors, service providers, and software companies, and found that so far, they're not experiencing a mainframe-resource drain. Says Cohen, "They have people on the bench, ready to go."
One mainframe user has turned to a storage solution for year 2000 to spare its production mainframe resources. Lilco used outside consultants for the application-data changes, and it's now using an EMC Corp. storage solution for the system testing. "We looked for a solution separate from our production environment, and EMC had the best offering to build and rebuild as we needed," says Steve Trezza, Lilco's storage-management supervisor.
Slick Trick
Using both EMC hardware and software, Lilco can test and then use production data to refresh the test data, all while the system is running. "EMC gives us a slick way to clip the data and move it to begin testing," says Lilco's Langton. "After testing, we don't have to delete the data or dump data out to tape for the next test iteration. We copy it over while our other work continues."
Companies resort to other measures to spare production resources. Some pay as much as $30,000 a day, plus travel and lodging, to business-recovery vendors such as Comdisco Inc. in Rosemont, Ill., for a testing environment. Others rent systems from Comdisco or grab a Time Machine 2000 mainframe from Amdahl for free (as long as they purchase consulting services from Amdahl's DMR consulting group). And many companies either buy dedicated Mips or partition existing systems after purchasing more CPUs. Amdahl's Foley says a dedicated system, even a rental, is the more secure move. "The partition puts more workload on the production machine and degrades the performance," he explains. "When an IT shop wants to stay in business, performance matters."
Glodo of Merrill Lynch says the personnel and system-resource problems won't go away with the arrival of the new millennium. "As people and resources become available after the year 2000, we'll have latent demand from projects that got put on hold," says Glodo. "Things will get interesting, because it's not like we win the year 2000 war and quit."
-with additional reporting by Bob Violino
Copyright (c) 1998 CMP Media Inc. |
