Brain drain looms

A mainframe skills shortage is emerging as subtly as grey hair. It’s bumping up training costs and raising concerns among data centre managers who wonder how they will replace retiring green-screen wizards with workers weaned on Windows and open systems.

Addressing a problem it says has the potential to become acute over the next five to seven years, a U.S. data centre professional association this month announced a program to help companies attract and train IT professionals to prevent a future shortage of mainframe workers.

At its semi-annual conference in Las Vegas, the Association for Computer Operations Management (AFCOM) outlined a data centre knowledge initiative that includes offering online courses in data centre skills in conjunction with Poughkeepsie, N.Y.-based Marist College and developing a best practice knowledge base to which data centre managers can contribute their expertise.

“We believe it’s a very important issue,” said Brian Koma, vice-president of marketing at Orange, Calif.-based AFCOM. He said the organization hopes the initiative will prompt IT managers “to take some early action” to plan for retirements “so they don’t have to spend more money later.”

In addition to offering hands-on training sessions at AFCOM’s conferences, the initiative provides for undergraduate and certificate courses in data centre technology.

Aging Mainframers

A study last year by Meta Group Inc. in Stamford, Conn., found that 55 per cent of IT workers with mainframe experience are over the age of 50. Conference attendees, such as Gerald Tucker, data centre operations manager at Foster Farms Inc., one of the largest poultry operations in the U.S., readily agreed with that finding. But he said he isn’t sure what to do about it.

Tucker has two mainframe operators with more than 20 years of experience who plan to retire in about six years, and he said finding replacements could be a problem.

“The solution could be an outsourcing possibility at that time,” he said.

The Livingston, Calif.-based company needs to find people with a rare set of characteristics: They must have good technical skills and be comfortable dealing with repetitive and mundane tasks, said Tucker. “They are usually one or the other,” he said.

Getting IT professionals, especially young ones, interested in learning mainframe work isn’t easy. But Ruben Trujillo, a technical specialist at U.S. Foodservice in Phoenix, said a company program seems to be helping.

New IT employees who may be destined to work in finance, for instance, can volunteer to tour the data centre and be exposed to its operations. “It’s opened up interest,” he said.

One of the data centre’s selling points is its critical, round-the-clock nature, said Trujillo. “The pressure and importance of it are very high,” he said, noting that that can be appealing to some prospective employees. Moreover, the work schedule – 12.5-hour shifts with alternate three- and four-day weeks – is attractive to some, said Trujillo. “I love it. It allows me to spend more time with my family,” he said.

Training Costs Rise

IT managers said they’re experiencing a rise in training costs. Mainframe skills aren’t widely taught, which means companies typically have to pay a premium for travel and training that can be as high as US$30,000 to US$50,000 annually per employee. “Our training budgets are definitely going up,” said Gary Rose, information services operations manager for Bexar County, in San Antonio, Tex.

At the same time, many companies are confronting the mainframe training issue in reverse, as they begin to phase out mainframe systems.

For example, Mattel Toys Inc. is training mainframe operators to work on Sun Microsystems Inc.’s Solaris operating system, among others.

Learning these systems isn’t easy for mainframe workers, said Lynn Hartman, data centre manager at Mattel in Phoenix, Ariz.

“For the older guys, it’s a major leap,” Hartman said. “But we made it clear it’s a necessity, not an option.”

The rush to consolidate servers and the data centre is on. Many companies appear to be squeezing redundant hardware, software, maintenance and service costs out of their data centres – and the effort is quickly paying off, according to IT managers.

Bayer Corp. in Pittsburgh offers a dramatic example of how much money can be saved: The company last year consolidated 42 data centres into two and cut the number of servers from 1,335 to 615. Expected savings: US$76 million over five years, according to Hobart Moore, an IT manager involved in the project.

The initiative was completed in July and delivered US$10 million in savings last year alone, said Moore, who spoke about the consolidation at a data centre conference held here last week by the data center professional association AFCOM.

“We showed that we aren’t just out there providing services and that we can actually save the company money when it became necessary to do so,” said Moore. “We came up with very solid ways of reducing cost without slashing services.”

In fact, many IT managers at the conference said they are involved in data center or server consolidation projects.

Over the past two years, Williams-Sonoma Inc. has purchased five of IBM’s high-end Regatta Unix servers at a cost of $1.5 million each. The San Francisco company is halfway through a year-long plan to reduce 100 stand-alone servers to five. Those servers will be moved to a partitioned environment on the Regatta servers, because partitioned servers can run multiple applications on the same box.

“You spend a lot of money upfront,” said Steve Stewart, the manager of the retailer’s data centre. But, he added, “it was either that or they were going to buy another 50 servers in the next year.”

Stewart expects a return on investment in two or three years and foresees savings coming from just about every aspect of the data center’s operation, including power, maintenance, personnel and licensing fees.

Eliminating one stand-alone server often allows companies to get rid of four more servers: the production server, the testing server, the fail-over server and a development server, said Marvin Hamann, director of operations at Supervalu Inc., an Eden Prairie, Minn.-based wholesaler and distributor of food and general merchandise to grocery and discount stores.

Hamann is reducing the number of stand-alone servers at his company from 120 to 60 by moving to partitioned servers. And that’s not the end of the consolidation. “When we get to that goal, we will set a new goal,” he said.

Consolidating data centres or servers isn’t a new concept, and IT managers at the conference were clearly aware of the potential problems and costs of allowing business units to add server after server to host applications.

But the continuing economic downturn has put pressure on IT departments to cut back. And, as Moore put it, “costs were just getting out of control.”

Mark Levin, an analyst at Meta Group Inc. in Stamford, Conn., called consolidation the “hottest thing” and said the most difficult aspect of the effort often is coming to terms with the inventory. Consolidating servers can lead to some management resistance, particularly if a business unit hosts a particular server. But as long as service delivery is unaffected, company buy-in is usually easy to get, IT managers said.

Michael McShane, manager of technical services and operations at Krasdale Food Inc., a White Plains, N.Y.-based food distributor, said getting cooperation from business units for consolidation projects is ultimately not a problem. “The dollars and cents of it convinces higher-ups, and it’s a done deal,” said McShane.