The 5 Keys for Business Computer Technology
Putting together the right set of technologies for a company is a very
confusing process. Vendors promise solutions which may solve one
problem, but exacerbate another. Islands of isolated technologies pop up
which enable a process in one part of the company, but cannot be integrated into
other processes in the company. IT people often seem overworked, and in the midst of continuous fire-drills
fixing the same problems over and over again.
A well functioning company has systems that evolve with the changing business
climate. The computer systems stay up. Different systems interact
with one another. And it doesn' cost gobs of money to maintain. And
IT people are able to work on improving systems instead of just struggling to
keep them running.
The 5 keys below are just a few of the guiding principles for computer
systems. There are others which may get described in a later set of papers, as
well as a technology set which solves most of the common business problems, using
standard technologies, which are available at
low cost, or in some cases, free. The goal is a minimum of maintenance, a
minimum of IT personnel, and an appropriate amount of functionality.
Communication occurs using
standard protocols. Proprietary software is avoided unless widely
available at low cost.
The 5 Keys for Business
Use low maintenance systems. A low maintenance system just
works, and keeps on working for years. A system which crashes every week
requires lots of IT energy. An organization will end up with dozens, perhaps hundreds
of systems running on its internal computers. If every crash results in
hours of work for an IT person, 20 systems alone will occupy one person full
time just to keep the systems up. This is very expensive, not only in
maintenance people, but in lost work time by employees.
Plan for interoperability. Not all systems have to work with each
other, but many do. Each time you double the number of systems, your
quadruple the number of potential interoperability points. A strategic
plan has to be drawn up to deal with this issue. You cannot build
automated processes requiring multiple systems without interoperability.
Use clear, easy-to-use, user interfaces. A system that is not easy to use may as
not be there. A system that is not easy-to-use will require excessive
training, and be error prone. Easy-to-use does not mean lots of
features. In fact, an excess of features is that biggest cause of
hard-to-use interfaces. Your feature set should match what you need to do,
and not anything more.
Don't sell your existing technology short. Every day some vendor
will come in and tell you why your systems aren't doing enough for you.
List to them, but always have a healthy degree of skepticism. Does
Office/XP really offer advantages to your company over Office/2000? Why
should you pay for an upgrade you don't need? Do you really need to switch
your website to use J2EE? It's working fine as it is. Is a 2.0Ghz
Pentium really more useful to your secretary than 233Mhz system? Most
people can answer no to all these questions, although there are a few with good
reasons to switch. Think before you leap into new technology.
Keep it simple. The final key is keeping it simple, and doing just enough to satisfy current
needs. Business systems evolve; they do not just get put into place, and
start working day one. A good business system starts out with humans doing
most of the work. Over time, the business process is refined to eliminate the error-prone or
repetitive jobs, and then if
cost-effective, automate the others.
A corollary of simplicity is not to over-design. Over-design comes at a price you may not be willing to pay, or even need to pay.
many web sites have been built with the idea that they need to handle the
traffic of Yahoo, when they only get 25,000 hits at day, something that
can be handled by a 486 based PC. How many Oracle databases have been
installed simply to handle a list of 500 customers? When you grow big
enough, re-factoring a system to scale it to a larger size is a nice growth problem
to have. You probably don't need scalability day one, and you probably don't understand your business
needs well enough on day one to warrant paying for a highly scalable