July 13, 2005
Health Care CIO Sees All CIO Roles Changing
By Evan Schuman, Ziff Davis Internet
Updated: Embracing custom coding—while evaluating VOIP and rejecting Linux and Oracle as too risky—reflects some of those modern-day changes.
As the CIO for a $4.1 billion health care products distribution company, Jim Harding knows only too well the changing face of medical technology today.
He sees senior physicians hand-writing prescriptions, penciling notes in patient folders and resisting electronics while the younger physicians embrace PDAs and digital transcription units.
But Harding sees the identical trend impacting IT management far outside health care, with younger office workers coming in with technology know-how that far outdistances their older counterparts.
"That's why the job of the CIO is much different today than it was 20 years ago," said the technology chief for Henry Schein Inc., the Melville, N.Y.-based firm that employs about 10,000 people in 19 countries.
"The user base is so much more educated on computers. They have a much better understanding of memory and disk issues. They know what a T1 is."
But as modern a workforce as he has, the technology Harding prefers to store and manage 50TB of data is decidedly old school: a group of AS/400 servers that he speaks of with mainframe-era admiration.
Referring to the legacy machines that IBM now calls the iSeries, Harding touts their reliability.
"They rarely ever—do I even know of a time?—they just don't go down, and they require very little maintenance," he said. "These machines are workhorses. The downside, though, is that you're dealing with IBM."
The heart of the company's data-crunching efforts falls to a quartet of four AS/400s that serve not only as a primary technology platform, but also a built-in place to consolidate functions that might otherwise become a proliferation of servers for e-mail, databases and other narrow functions.
"Everybody that I know is looking at server consolidation all the time," Harding said. "Is it truly better to buy 200 servers?"
Beyond the AS/400s, the company's hardware includes desktops from Dell and IBM, IBM laptops, Toshiba tablet PCs and about 250 BlackBerries (Research In Motion).
On the operating system side, the company has standardized on Windows XP and Windows 2000, but may become even more purely Big Blue by moving all Web efforts onto IBM's Websphere, because Websphere "has a lot more stuff out of the box than we can get in the XP world."
The company's approach with its AS/400s is certainly not unheard of.
"It's a question of having a single point of functionality instead of managing an entire environment of 20 or 230 servers," said Mike Kahn, managing director of the Clipper Group Inc. analyst firm. "The partitions are strongly separated in the I-series, allowing better security. So, if there's a problem, it's isolated."
Mike Chuba, a vice president with the Gartner Group of Gartner Inc., agreed, adding that—even today—the AS/400s can provide some unique functionality. "There are reasons why some go with platforms that are the more expensive, bigger ones," Chuba said. "Error-correcting code could be built into a more expensive system. That isn't there in a simple, more commodity-designed system."
The AS/400 installed base also has a very strong incentive to stay put.
"The cost of migration is a big factor that keeps people where they are, as well as legacy workloads. To go to a commodity designed system—it's a massive decision process," Chuba said. "If you move over from a platform with five series lines of availability for some platform that has lower standards of availability, that could be expensive. It can make strong economic sense to keep a system where it is."
Harding is also a strong proponent of extensive application customization. Roughly half of all code on his network is homegrown, he said.
He described his fondness for custom software as much less of a financial issue—although that plays a role—and more as being because he frequently needs very specialized functionality quickly and it's simply faster to build it than find it.
"If we need to enable a new business process, a new pricing scheme, a new chargeback system for our vendors, [with custom coding] we're able to get it done in relatively speedy terms," Harding said. "We don't have to wait for a package or a release, which means that we can react very quickly to a business need. I feel very good about the flexibility there."
The Clipper Group's Kahn said that custom coding is becoming more attractive.
"A lot of people feel that custom coding gives them a proprietary or strategic advantage. Each company has to determine if that's the advantage or edge they need. Some companies have to have that something that makes them different," Kahn said.
"You can't generalize that. A system that works at one hospital might not work at another one. Sometimes companies buy these pre-packaged solutions only to find out they can't tolerate it," he added.
The nature of Henry Schein Inc.'s operations is twofold: Harding's team must support those 10,000 employees globally, but he must also package the technology to be resold to about 475,000 physician offices, dentists, laboratories, government facilities and veterinary clinics.
Those two roles have very different responsibilities, and Harding is preparing to split those roles into two positions, with Harding keeping the customer part for himself.
That resale role is another key reason for his fondness for custom code, he said. It provides him much more flexibility when it comes to pricing, for example.
In some sales situations, the company would like to provide a certain pricing approach, but application licensing rules sometimes lock him into pricing that makes sense for neither his company nor his company's customer.
"We have to say, 'You can't price that way because that's not the way SAP priced it,'" he said. With homegrown apps, the company has far more flexibility to price them as business needs dictate.
That's not to say that Henry Schein Inc. avoids using off-the-shelf apps, but it is doing a higher percentage of custom coding—about 50 percent—than is typical of other companies its size. Some of the company's larger applications are from Siebel, SAP, JD Edwards, MicroStrategy and Informatica.
"Those software packages "are generic. We have to make them fit for us. Our strategy has been to have a package foundation and build a lot of custom around it," Harding said.
"If I could buy it and it was clean and off-the-shelf, we'd do it, but we have to modify."
He cites as an example the company's challenging warehousing situation. "Supply chain is the heart of our business. We have more than 100,000 SKUs. We have items as small as the tip of your finger all the way up to very large equipment," Harding said. "Any order we receive by 5 o'clock, we're shipping that day, so any interruption, any slowness, will have a major effect in a hurry."
A traditional buy-versus-build argument in favor of buy is that custom code needs more staff support because there's no outside vendor-provided helpdesk that could assist. But Harding found that argument didn't work for Henry Schein Inc.
"We have a small team on warehouse: four people. Even if we had a package, we would still need people for support, for interfacing," Harding said.
"It was much pricier for us to roll out a point solution versus what it would cost to enhance the system we have."
Before he worked for Henry Schein Inc., Harding spent 20 years with Mobil Oil in a wide range of capacities, and that background has influenced him to take a broad view of ROI (return on investment) and TCO (total cost of ownership).
At Mobil, Harding "worked as a business person trying to make things happen" and was a divisional CIO and held senior management roles in operations, marketing, financial/corporate planning and "everything but sales."
That's why he stresses that his attraction to custom code has relatively little to do with the initial cost savings. "At the end of the day, the cost wasn't the major driver. It was more about speed and 'How do we stay ahead of the competition?'" he said.
"But the [custom] costs were indeed reasonable [compared with] the licensing fees and the maintenance fees, and you still need people to understand it."
Another typical concern raised about custom coding is long-term support. When the people who created it retire or resign, will there be a cost-effective way to bring people in to keep it working?
"I think it's overplayed how big a problem [long-term support] is. I've just found that there are people out there to support these custom systems, although they can sometimes be hard to find," Harding said. "I've never run into a situation where [custom] prevented me from doing something" that involved integrating with other applications, he said, referring to UPS and Kmart as other large shops that use a large amount of custom code.
Despite recruitment fears, he found that searching for talent to support custom, legacy and even—a few years ago—Y2K was quite easy.
But he did find difficulties in the late 90s when Web development was at its height and programmers were being gobbled by tons of E-Commerce startups.
"The worst crunch [historically] has had nothing to do with legacy [or custom] code. It was the dot-com boom when everyone was trying to replace everything," and that's when finding good code-crunchers at reasonable prices was difficult, Harding said.
Harding predicts that custom code creation and support is going to be even easier in the coming years, and he thanks the Internet's popularity for that. "If you look at the Web world, that whole world is essentially custom," he said.
Harding said that his reluctance to give too much weight to price also impacts his feelings on two technologies that have been propelled to stardom based overwhelmingly on their perceived cost savings: Linux and VOIP (voice over IP).
As for Linux, Harding said, "It's not high on our priority list. We haven't seen the benefits of going that way."
He added that this is despite the fact that an opportunity to get away from Microsoft is "a good driver" for Linux, as is the fact that Harding has seen some of their applications—such as a key tax package—become no longer available on the Windows platform. "It won't come up on Windows. Just Linux, Unix and no Windows."
Despite that, he said, he is simply too afraid to place mission-critical applications on Linux: "Put our core apps on Linux? I couldn't even think of taking that risk."
Risk aversion is one of Harding's central themes, and he also mentioned it when discussing concerns about the latest rash of Oracle acquisitions, including JD Edwards, PeopleSoft, Retek and ProfitLogic.
He said he wants to expand his use of SAP in Europe and he wants to bring a lot more of it into the United States, all mostly because of his fears of where Oracle may be heading.
"My fear is J.D. Edwards, because that's where we have more risk. How long is it going to be supported?" he asked. "I am very concerned about that. That puts me in a barrel. That's why we are moving to SAP overseas and preparing to bring that into the States."
He added that this is all about "future-proofing."
"To make a major multimillion-dollar decision here on a new ERP system, I have a hard time picking Oracle because I don't know where it's headed," Harding said. "Are we going to have to migrate? It's a very dicey world. It's much easier to pick SAP because you know what you're getting. Would I want to stand in front of the chairman and put our company at risk like that? Not unless I absolutely have to."
Oracle executives have argued that SAP's future strategy is just as much in question, as they are rumored to be still searching for other companies to acquire.
But VOIP is a very different situation. Henry Schein Inc. spends more than $10 million a year on its phone bills, mostly because of its large call centers.
"We're looking to make a major cut in that," he said.
Harding has been watching VOIP for many years, but today's implementations are using regular analog phones and earlier problems with latency are now said to be minimized.
"VOIP has been around for years, with broken promises. I'm not going to have voice over IP because the chairman read about it in a magazine. We really want this to work," he said. "The difference now is that the Internet makes it possibly more viable. It's started to get some real play, but whether we can do it in the enterprise, I just don't know."
Harding is investing about $150,000 to doing a small test of VOIP, which will include some Cisco equipment.
"We're just making a calculated risk on the investment," he said. "Any change requires investment and new technology normally requires a substantial level of investment."
But Harding stressed that telecommunication systems are mission-critical and that anything less than an outstanding performance—no dropped calls, no employee complaints about line audio quality—will kill the project.
"If the test isn't dang near perfect, it will be a show-stopper," he said.
Editor's Note: This story was updated to include analysts' comments.