David Strom

Planning your network upgrade

By David Strom

What do using technology and running a marathon have in common? Even after lots of

training and conditioning, you still can find obstacles around every corner, not to mention

"hitting the wall" midway through the process and finding out you've picked the wrong set of

computing products to run your business.

How can you plan adequately so that you won't end up with stuff that is obsolete in just a

few months? And since the rate of change for computing technology is furious, how can you

keep abreast of these changes without turning it into a full-time job reading the trade

magazines? Finally, can your business get by without having to hire a staff of information

systems (IS) specialists?

Unlike preparing for long-distance runs, there isn't any simple high-carbohydrate diet that will

keep you alert enough to technology trends. But before you give up hope of any regular

exercise and turn into a computing version of a couch potato, there is hope. Here is our recipe

for success.

1. First, let's look at how you should buy your desktop hardware. You should walk a narrow

line between avoiding the brand-new and sticking too long with ancient equipment just

because it is cheap. One strategy is to buy gear that is between three to twelve months old

and based on a processor that is just shy of the currently fastest available. Processor speed is

measured in megahertz: the higher this number the faster the processor. For example, the

current speed champions are the 60 MHz Intel Pentium used in IBM, Compaq and other DOS

and Windows PCs and the 33 MHz Motorola 68040s used in Apple Macintoshes. 

This strategy gives time for the initial buyers to find and work out any problems, and usually

to take advantage of the first round of price cuts that are inevitable in this business. It also

helps to have a few months' perspective after a new product is shipping to distinguish

between what is new and what is merely repackaged to look new. This is especially true for

volatile markets such as notebook computers, where new versions with minor variations come

out almost monthly. And the strategy ensures that you are always purchasing relatively new

gear thus staying near the top of the technological learning curve, but not at the bleeding


"We are technology followers: we don't want to be the first ones to buy a new Pentium. We

try not to overpay for being on the cutting edge," says Ed Fischer, manager of computing for

the Tribune Broadcasting Co. in New York, NY (212 210 2588) Tribune has learned its

lesson: in the past they were very conservative and purchased machines that were several

versions behind the times. "We bought 386SX machines when the faster 386DX processors

were available for $500 more. That was a mistake: as we move towards Windows, the SXs

are just too slow," says Fischer. As a result, they got saddled with a large supply of older

machines. Now they are more aggressive, but are still staying away from being first for new

products such as local bus video.

A similar situation happened at a major California defense contractor, who purchased too

many 286s for too long when faster technology was available. "We only pick the latest

technology when all the initial bugs have been shaken out," says Tom Simondi, who is

responsible for corporate computing guidelines there. (805 734 8232 x 60092) He usually

waits until a second version of the technology is available, or when the initial rush of

problem reports has faded from the trade press. 

2. Next, know when to upgrade your equipment and keep depreciation cycles to as short as

two years.  "Depreciating equipment any longer than two years wastes money and encourages

companies to keep ancient gear around. The accounting practices should match the life of the

equipment," says one information systems executive that has managed to convince her bosses

of the shorter depreciation cycles. Convincing the Internal Revenue Service is another matter,

where longer cycles are still the norm.

Knowing when to upgrade can be a black art: one suggestion is to only jump every other

generation on hardware, such as from a 286 processor to a 486 processor, or from a 386 to a

Pentium.  "If you change every generation, you'll be in a constant learning mode and never

really see the productivity boost you might otherwise by using a combination of hardware and

software for a longer period," says Simondi.

Another suggestion is from Tribune's Fischer:  "Try to look at the value of the additional

speed, features, whatever. When they become so clear cut, you should make the jump." Some

examples that can force an upgrade: when you move from DOS to Windows, you'll need a

more capable machine. Or when moving from stand-alone to network attachment, a different

configuration is usually called for. 

3. Third, only upgrade software at significant revisions, and upgrade everyone at once in your

organization so that they are all running the same software version.  This is especially true for

networked systems or in organizations of less than 200 employees. Figuring out what is

"significant" can often be a judgment call, but one way to judge is when you start seeing

many new advertisements from the software vendor promoting this new product revision.

"Novell has been telling customers for almost three years to move to new client network

drivers. [the software that Novell supplies that matches the particular network adapter card in

each PC] Take that hint, and upgrade clients when you have the chance. Make a conscious

decision that every new client from this day forward will be configured for the future," says

James Gaskin, president of his own Dallas-based consulting firm, Gaskin Computing Services

(214 222 0900) In other words, use networking software that will have legs and work with

upcoming products and enhancements such as Novell's Directory Services, even if you don't

necessarily require them for today's operations. Gaskin goes in and does these network driver

upgrades often with his consulting clients, since they usually fix a number of other problems

as well as keeping his clients' equipment current.

4. Fourth, keep up with business changes. If you are adding or subtracting staff, acquiring a

new division, or introducing a new product line, it is time to re-examine your technology and

applications choices. Does your accounting system match the firm's current needs and size?

As a result of a merger with another company do you have two electronic or voice mail

systems that can't communicate?  Is data being input manually more than once into two

different systems? "No one can blame customers for the choice they made five years ago,"

says Gaskin. "What you can blame them for is not checking that decision to make sure it is

still valid." 

He recommends an annual technology audit to examine how close technology matches these

business processes and to shed any light on these computing dark ages. "One customer of

mine sells reports that involve intensive number crunching." The time involved in producing

these reports is critical. Gaskin convinced them to upgrade an aging Pick System to an IBM

RS/6000. As a result, they improved their performance five-fold measured by the time needed

to complete each report. "After the upgrade, they discovered a mistake in one of their

equations and had to recalculate everything from scratch. With their old system, it would have

been impossible and they would have lost hundreds of thousands of revenue dollars. This

mistake could have ruined the company." So the upgrade, which was done for performance

reasons, saved them when it came time to redo their mistakes.

Ideally, your applications needs should keep up with your changes in your business, but

usually these applications are rarely top of mind: "We didn't think about our applications. As

our staff has grown, we are outgrowing our applications. Now we need new applications in

addition to the hardware and operating systems to support them," says Holland Carney.

Carney is a senior consultant for Alexander Communications in San Francisco, CA (415 923

1660), a hi-tech public relations firm with offices in Atlanta, Ga.

"If we had to do things differently, we would put the effort into projecting our applications

needs five years in the future."

Carney mentions her firm's use of electronic mail and databases as two examples of

outgrowing her applications: "We started out using CE Software's QuickMail, (Des Moines,

IA  800-523-7638; 515-221-1801) but some of our managers process over 100 messages

daily. We found that QuickMail just isn't robust enough to handle our needs, especially when

it comes to gateways to public systems such as MCI Mail."  (MCI Communications, McLean,

VA  800-333-1000; 703-506-6000) Carney is looking into Lotus Development Corp.'s Notes

(Cambridge, MA  800-343-5414; 617-577-8500) software and other systems that can provide

more firepower.

"We also use FileMaker Pro on the Mac (Claris Corp., Santa Clara, CA 800-544-8554;

408-987-7000) to keep track of our contacts. Now we have multiple databases running, and

any changes have to be made in every version. It's time for a quantum leap to a powerhouse

database like Paradox," (Borland International Inc., Scotts Valley, CA 800-682-9299;

408-438-8400) she also said. 

5. Fifth, invest in your cable plant, especially if you are moving into new facilities.  "Ideally,

you'd like your cabling to last twenty years," said Fischer, "but if you asked me even five

years ago to predict what kind of cabling we'd be using, I wouldn't have made the right call."

"When we had 12 machines to network, LocalTalk and Macs seemed the cheapest and easiest

and didn't require much training. Now we have outgrown this and so we have planned ahead

and wired our offices to support Ethernet," says Carney. They could have saved themselves

from redoing the wiring had they planned ahead and put in Ethernet from the beginning.

6. Sixth, know your own limitations and when it is time to get help. Your choices are to go to

your peers, get additional staff, or hire consultants.

The best peer groups are those run by various local chapters of the Microcomputer Managers'

Association (the national office is in New York and can be reached at 212 787 1122). "Go to

their meetings and chat with people who appear to be in similar situations," said Simondi.

Besides these local chapters, computer user groups exist in most major cities as well.

Members are drawn from small to large businesses, and include a wide range of professionals

who manage their microcomputing strategies and applications.

Know when it is time to hire a full-time information systems (IS) manager. Things to look for

that make a full-time IS manager critical: more than one office, a high percentage of traveling

users, and several computing applications that are critical to business operations. "In

hindsight, we should have bitten the bullet and hired a full-time IS person sooner," says

Carney. Many of their existing applications were more hobbies that grew out of personal

interest than part of anyone's job description. "Right now the IS function is distributed among

two or three power users -- it is no single person's responsibility. We had a part-time support

technician for awhile, but all he did was put out fires. This model doesn't work for 30-plus

users in two separate offices and notebook users traveling all over the world."

If you can't gain access to your peers and don't want to increase your staff, it may be time to

look for a consultant. But finding the right one, or one you can trust, may be difficult.

"If you don't have a technology background, hiring a consultant could be dangerous: you may

not know until it's too late that you've been had," says Simondi. "The best situation is to hire

two people: one to recommend a course of action and one to implement it.  This minimizes

conflict of interest problems."

"If a doctor has done lots of ulcer operations, and most of the patients live, that's a good

reference. Unfortunately, computer systems aren't as tidy and consistent as stomachs," said

Gaskin.  Nevertheless, he recommends that you look for someone that can listen to your

specific requests and respond accordingly when they write their proposals, rather than selling

you last week's work. "Remember that people with a hardware inventory can't tell the whole

truth. When someone has to choose between your project and their profit margin, you will

lose every time."

"It's impossible in this game to avoid some hard knocks. No matter what you choose, you

will encounter unanticipated problems," said Simondi.


David Strom is a frequent contributor to Forbes ASAP and has his own consulting practice in

Port Washington, N.Y.

Click here to return to the previous page

David Strom David Strom Port Washington, NY 11050 USA US TEL: 1 (516) 944-3407