News that a network is down can reverberate through an organisation with the speed of a "free beer in the kitchen!" announcement over the PA system. Prolonged downtime can put enough red ink on a profit-and-loss statement to convince even the stingiest bean-counter that maintenance agreements are a necessary cost of doing business. It's like health insurance: it might not be needed very often, but it pays for itself many times over when disaster strikes.
However, the days of simply purchasing a maintenance contract as an add-on cost to a PC or ser-ver are over. The typical support agreement, in which the manufacturer directly provides on-site service and support, has been replaced with a booming service industry that isn't just interested in maintaining PCs and printers. Instead, it wants to support the entire network infrastructure.
Manufacturers have traditionally provided on-site service and support for systems and peripherals, but they are increasingly off-loading this business to third-party service companies. However, even these companies are loath to take on contracts without getting a bigger piece of the support pie.
Almost every piece of hardware comes with a multiyear warranty. If you read between the lines, the vendor will usually swap out failed components on a 24-hour replacement basis only during normal business hours. Even if the server or router goes down at 5pm on Friday and the business operates around the clock, seven days per week, the replacement parts won't arrive until the following Tuesday.
Most IS shops insist that hardware failures and disabled networks be repaired within specific time limits - usually between 1 hour and 4 hours - no matter what time of day they call.
Vendors have offered such extended service warranties for a number of years, but their days are numbered, says Eric Rocco, a senior analyst at market research company Dataquest. "There is a definite trend among manufacturers to get out of the on-site service business on their low-end products, like PCs and printers," Rocco says.
And they're subcontracting that process to their reseller partners or to nationwide service organisations. That's even becoming true of more critical components, such as servers and mass storage subsystems.
But large service organisations aren't jumping through hoops to get a lot of PC and printer repair business. It isn't profitable. They want to service the entire network infrastructure and the desktops as a complete package.
Analysts see this as a growing trend within the industry, and it is shaping up as a huge chunk of business. Although a few vendors still find extended maintenance agreements profitable on big-ticket items, many are off-loading their contracts to independent service organisations.
"You see a lot of this type of subcontracting going on today because no one, not even a com-pany the size of IBM, can possibly service every one of its own products on a priority basis," says Doug Chandler, a senior analyst with IDC.
But having dozens of separate high-priority agreements isn't a practical approach. You don't want six or eight technicians from many vendors pointing their fingers at each other as they try to resolve a major networking problem.
"The large percentage of problem calls we receive on our communications hubs pertain to interoperability issues," says Steve Ruzinski of US Robotics.
Analysts say this makes a strong case for having only one or two companies service your entire network infrastructure.
"It makes sense for us," said one government buyer. He said he has problems primarily in remote and rural areas. "If a service company can spread its risk over a large piece of business, they can afford to give us all the attention we need."
Other large organisations or businesses say that a single vendor may have breadth of coverage but not always the depth of experience to cope with every disaster it may be called upon to solve. IS managers and analysts advise companies to check the service provider's qualifications and its track record with other clients.
Worth the cost
Organisations that moved to distributed computing understand the value of a maintenance agreement. "We automatically contracted for maintenance services as soon as we began to install our LANs and WANs," says Nick Muraca, a network manager. Muraca once prepared a study to determine the business loss if 150 users on a server suddenly found themselves idle.
"The answer was simple to obtain," Muraca says. "We added together the salaries, benefits, utility costs, the volume of business we conduct on average per hour, and the sheer waste of employee time. And upper management agreed that a service contract was not a debatable topic."
Yet there are companies that shun maintenance agreements like they would a pickpocket. "I can't convince our financial officer that maintenance contracts are necessary," said an IDG reader who requested anonymity. "He has this notion that once you buy something, it should work forever."
"I shudder to think what would happen if our Internet router, firewall, or Web server crashed. None are on maintenance, but the CFO screams if anything is down for more than an hour," the reader concluded.
This reader's employer is the exception rather than the rule. IT executives overwhelmingly say their senior management understands the need for some kind of comprehensive service program for the mission-critical hardware in an enterprise.
"You can't be penny wise and dollar foolish," says IDC's Chandler. "Systems are too complex, too interdependent with one another. A network failure doesn't always occur because a box breaks."
The root of the problem could be something as simple as an improperly configured server or even a software application.
The median cost of network downtime, according to Dataquest, is $US4,000 per hour. In some industries, such as finance, transportation, manufacturing, and retailing, that cost can easily climb to $US100,000 per hour.
There are no standard maintenance plans. Each service aggregator has a portfolio of service options with varying degrees of priority. Customers pick and choose from line items as they would from a Chinese takeaway menu. They can have anything they want - from 1- to 2-hour priority response to 24-hour service, and they pay accordingly. Even the total price of an agreement is then subject to negotiation based on its length of term and the additional service the customer wants.
The larger national service providers are also pushing extras such as on-site engineers and network management to companies that have downsized their IT staffs. And because service companies have relationships with all the major equipment manufacturers, they will assume any separate priority response agreements users may have with vendors.
The rates service companies charge are prepared along similar lines to the actuarial tables insurance companies use, according to one large service provider. "We have to make a profit to stay in business, and our rates on specific components and platforms will change according to the incident rates." But he added that his company's fees include pro-active, preventative maintenance that fine-tunes and synchronises network components, as well as microcode and software updates and even training classes for IT staffers. If the incident rates drop, so will the prices of their contracts in subsequent years.
Some companies that don't regard their computer systems as mission critical have been opting for a "pay as things break" type contract. But Tom Sweeny, research director for Dataquest's software services program, says: "Interest in per-incident type contracts is waning because companies aren't convinced they derive greater value when they pay as things break. There are just too many highly critical components in place today."
Corporate comfort levels vary considerably, Sweeny says, and he points out that certain industries, such as finance, transportation, and retailing, look at their entire infrastructure as mission critical and are buying service contracts to keep all of their systems up and running - at least to the extent that they need to drive their business.
Organisations with strong in-house technical capability might choose to service specific hardware their own staff understands and contract with a network specialist to cope with other segments of the infrastructure.
But even with a wide range of options, IS managers are likely to see higher maintenance costs as the service industry copes with its speedy growth. Industry insiders admit that there is a growing shortage of skilled labour in the industry, which is driving up salaries and maintenance costs.
Maintenance manager Susan Bailey says quite frankly that high-level engineers can write their own contracts today. "The demand for experienced talent is greater than the supply," Bailey says.
Features and benefits
IT executives say the biggest benefits they receive from maintenance agreements are the pro-active maintenance effort, system finetuning, and incident logs.
Is the service industry doing a good job on the whole?
Dataquest's Sweeny believes it is, though it is still a relatively young industry experiencing growing pains. The major players are rapidly gobbling up the smaller, struggling companies to tap into their account base and technical talent.
"They also realise that it's becoming a very competitive marketplace out there," Sweeny adds. "If they don't deliver quality service under their agreements, customers know there are good alternatives they can turn to."
Treat your service provider as a partner"People who want to keep networks humming have to think of their service provider as a partner, not just a bunch of guys with screwdrivers fixing things when they break," says Gerard Higgins, a US IT manager. Higgins places a strong emphasis on system availability by focusing on reducing the amount of downtime of each incident. This focus has resulted in big savings for his company.
During the past several years, the average duration of an outage has dropped from 107 minutes to 23 minutes.
This wasn't a simple task, because the company has one of the largest computing infrastructures in existence. It has 35 large-scale IBM mainframes at three central data centres; 700 minicomputers at five regional data centres; and 35,000 PCs across 80 locations.
Higgins says the IT workload has increased 400 per cent during the past few years, while he has had to reduce the IT staff to 900 engineers and technicians from a previous high of 1,800. But Higgins has still been able to accomplish these goals by working closely with his service providers.
"We place a dollar value on the financial impact of each outage," Higgins explains. He and the supervisors from his data centres and from the service companies sit down with the company's president every month to review the performance of each piece of equipment, every operating system, and every database system.
Although these meetings aren't always happy occasions, Higgins says there is a great spirit of cooperation between IBM and the other service companies because he offers them a tremendous incentive to keep reducing the downtime of each incident: "Contract renewal."