Page 206 - Accounting Best Practices

c10.qxd 7/31/03 3:12 PM Page 195
195
10–8 Improve Computer System Reliability
other’s functions. If one of these computers develops a problem, the other
one takes over all processing and continues operating on its own so that users
have no idea that there is a problem. The damaged computer can be repaired
while the other unit continues to operate. Though this is a more expensive
approach, it guarantees a very high level of system reliability.
• Emergency planning and testing. No matter how many precautions a com-
pany takes, it is likely that there will be system crashes from time to time.
Rather than passively hope that these incidents do not occur, it is better to
develop a formal plan for how to deal with them before they happen. By
writing down the precise recovery steps to be followed, one can save a sig-
nificant amount of time in fixing systems. This plan can also be used for
practice; by scheduling periodic training sessions for recovering from system
crashes, one can determine the weak points in the emergency plan, and fix
them before a real emergency occurs. By using this approach, a company can
keep system downtime to a minimum.
• Redo cabling. Some employees have difficulty staying on-line with their central
computer systems. This is caused by poorly constructed network cabling,
which may in turn be caused by excessive cable lengths without repeaters,
cables running near power sources (such as machinery), or the wrong types
of cabling. In some cases, the best way to eliminate this problem is to com-
pletely redo the cabling. This may require the installation of top-quality,
high-capacity fiber optic cabling, as well as new hubs. Also, if there are links
to distant locations, it may be necessary to convert from a dial-up modem
access, which runs on standard copper cabling, to a high-capacity T1 phone
line, which is much more reliable, although also much more expensive to
operate. By making these changes, a number of system reliability problems
can be eliminated.
• Scheduled downtime. One of the most common employee complaints regarding
system downtime is that maintenance occurs during regular business hours,
rather than at other times. When maintenance, such as system back-ups, test-
ing, or software upgrades, is going on, other users cannot access the system,
which keeps them from performing their jobs. To avoid this problem, it is
very important to cluster standard maintenance work together in a batch and
run it automatically during low-usage periods, such as late at night. Simi-
larly, any other system work that may bring the computer system down must
be carefully scheduled to match low-transaction periods during the work-
week, such as just before or after the regular working hours, or during the
lunch period. The best way to ensure that these times are properly scheduled
is to create a work schedule for the computer department that identifies well
in advance the periods when the system must be brought down, so employees
can be adequately prepared in advance for these periods, and so additional
planning can be done to ensure that the downtime periods are kept to an
absolute minimum.