Down to Brass Tacks
Deciding on standardization vs. best of breed is not easy.
October 7, 2002
Standardizing on Microsoft across the board helps Continental Grading Co. in Chicago do without IT personnel. Continental is a small company, with 30 employees, and all technical matters are handled by the firm's CFO, Tony Aukett. "I use Windows at home, so when it comes to doing administration, it's familiar to me," Aukett says. "I'm not a particular lover of Microsoft, but it makes life easier for a small business."
Continental uses Outlook for messaging, SQL Server for database management and Windows XP on the desktop, and Aukett points out that while Microsoft products are expensive, his company doesn't have to spend money integrating software from a number of vendors.
For a small environment, we see Aukett's point, but remember, TCO (total cost of ownership) is a key factor to consider. The best-of-breed approach might cost more up front, but often you'll save money in the long run by avoiding out-of-control licensing schemes.
Mid-Atlantic Funding, a real estate company in Chantilly, Va., tries to have it both ways, says CTO Joel DeClue. "We standardize in areas that require a lot of support--infrastructure, desktop services," he says. "But in certain projects where we need maybe some new process to really kick things up a notch, I might go best of breed."For Web applications, DeClue has fiddled with IBM's WebSphere and BEA's WebLogic, but he usually finds his way back to Microsoft's Internet Information Server. "From a business perspective, it's hard to get away from Microsoft," he says. "You can hire anyone to come in and work on Microsoft."
One lesson to take away from this is that IT decisions are made for many reasons; technology alone doesn't cut it. This is why we have expanded our focus to include the business requirements that drive the technology-selection process--to give you an understanding of the factors your business must bring to the table.
As for what you're running your business on, Microsoft's platforms are far and away the most popular among those surveyed. Respondents estimated that 63 percent of their server OSs are Windows, with the largest installed base among small businesses (one to 100 employees). Small businesses also have the highest installed base of Linux servers, not surprising when you consider the software's price (free) and the wide variety of hardware platforms it can run on.
Very large companies (10,000 or more employees) make up the largest percentage of installed Solaris and other Unix (not Linux) servers. Microsoft still has a large presence here, but this is where Scott McNealy and Larry Ellison like to earn their money. And with Sun recently announcing its new Linux server products, you can see where the company is looking to gain some server ground.
At Your Servicesome readers said they would love to standardize on Web services but must cope with multiple Web platforms. The frantic Web development cycles of the '90s, coupled with a rapidly changing vendor marketplace, led to these shops making decisions about which Web development environment would best serve their needs on a project-by-project basis. Now it's difficult to turn back. In other cases, the diverse Web environments can be traced to company consolidation. "A lot of the reason we have so many different Web environments is acquisitions," says James Chance, infrastructure specialist at EDS, a systems integrator in Plano, Texas.
An interesting point we discovered when we asked about Web services was the general confusion in the market. Most respondents listed their current Web server platforms under the same technology heading as Web services. As much as Microsoft would like to think that people understand what .Net is all about, from where we stand there is plenty of room left for education and market differentiation. One thing that is clear is that Sun's iPlanet strategy and Web services architectures from IBM and HP are fighting an uphill battle in the name recognition war with Microsoft. Market hype from the vendors and technology press (yes, us too) makes defining Web services quite a challenge for technologists--never mind the level of understanding, or lack thereof, of business people.
It's not always immediately clear to non-IT executives why standardizing on big-name vendors can be less expensive, especially when the purchase prices are higher. Execs at Progressive Life Center, a children's services organization in Washington, used to favor products with the lowest list price, says network administrator Vanessa Hill. Now IT is beginning to convince them that they can save money through easier administration, and more IT purchases are moving in that direction, Hill says.
In some ways, the movement away from best of breed has taken some of the fun out of technology purchases. "Standardization keeps the workflow going and is good for training purposes," says Paul Davis, IT manager at Rinehart Oil, in Ukiah, Calif. "A lot of times I'm itching to get something great and fresh and new that's out of the ordinary, but I haven't done that in a while."
Sure, cool is fun, but let's not lose sight of the fact that having a stable, reliable technology environment leaves IT more time to create long-term strategies and perform technology evaluations. Vendors are usually open to providing demos, so there's no reason to stop looking at new technology, whether you have the budget to buy it or not.
DuPont e-business architect Gil Choi calls the debate about standardization versus best of breed "a loaded question." Most times it may be best to standardize, but if you're too strict about it you may not be able to find technologies within your preferred list that meet your business needs. He also says CORBA and XML interfaces between apps eventually will allow for the best of both worlds: best-of-breed software that functions predictably like one common system. This is the promise of Web services.The State of Montana has a short list of standard vendors--Oracle for databases, IBM for servers, Dell for desktops. But it made an exception when it chose PeopleSoft to handle human resources, payroll and billing two years ago, and the shift proved costly. "It's been a good source of controversy" among IT staffers, says systems analyst and engineer Mike Hahm, who notes that the decision came from the top--another case where IT was not included at the start of a project but is now feeling the pain.
Despite economic uncertainty, many IT shops are forging ahead with infrastructure and enterprise application projects. Asked to rank their spending priorities for the next budget year on a scale of 1 to 5, respondents ranked infrastructure highest (44 percent gave it a 4 or 5), with business applications trailing slightly (43 percent). Security was also cited as a priority (42 percent). But don't expect new staff to help you with these projects: Spending on head count and consulting services ranked among the lowest priorities, with less than a quarter of respondents giving them a 4 or 5 rating. Of course, this disparity could be misleading because consulting and head count are funded out of operating budgets, while most IT projects are capital expenditures--one-shot deals with some long-term licensing and maintenance costs. Companies always want to decrease operating dollars, hence the reason bodies and training don't get funded.
Asked more specifically what their most important funded projects are, 34 percent said business applications, followed by storage and servers (15 percent), infrastructure (11 percent), network and systems management (10 percent) and security (9 percent). Asked to rank important projects that have no formal funding, business applications also led with 29 percent, followed by network and systems management (16 percent), security (12 percent), and storage and servers (11 percent). An effective tactic here is to look for opportunities to weave "unfunded" projects into funded ones. For example, with business applications, always make the case that support technologies--network/systems management, core infrastructure, security and storage--are critical to the successful rollout and upkeep of the application. The only way to make that case is to be involved when the project is getting the funding it needs--even though most companies plan for a 10 percent to 15 percent contingency on large-scale projects, consulting can swallow that in one gulp.
Davis of Rinehart Oil illustrates the difficulty of helping business applications evolve without funding. Rinehart's main accounting software runs on Unix, but everybody uses Windows on the desktop. The vendor of the accounting software will sell Davis translation software, but it's expensive and outside his budget. So he is forced to use a less elegant approach. "It's a matter of teaching people how to manually move data to and from Windows," he says. Ouch.
Davis' biggest budgeted project is to standardize desktops across the organization. PCs were bought piecemeal until recently. "When a workstation broke down, we went down to Staples and got the cheapest Compaq," he says. Shortsighted decisions were being made, and purchasers didn't consider the impact on consistency from the perspective of IT staff or even individual users.It's a case study on why TCO is so important. Standardization is typically looked upon negatively, as something IT is dictating to the user community. Instead, Davis should build a case why cost of ownership, for both software and hardware, goes through the roof when every desktop is different. These are hard dollars that are easy to quantify. There's also a soft dollar cost around productivity decreases and potential critical data loss. All these facts should be in your arsenal so you can approach problems from a business perspective instead of a traditional technology perspective. That way, everyone wins.
David Joachim is Network Computing's editor/business technology. James Hutchinson is Network Computing's director/editorial content. Write to them at [email protected] or [email protected].
You May Also Like