Nortel 400, Using Manual

Page: 209 / 344

Chapter 7 Web cache configuration 209
Using the BayStack Instant Internet Management Software Version 7.11
When you first install the Instant Internet unit, you should run it with the default
settings until the cache entries fill up to 100% (Cache entries % full statistic on the
Web Cache Statistics page) to establish a benchmark against which you can
measure future changes. After you review the statistics and understand the savings
you gained with the default settings, fine-tune the cache settings and begin your
experiments.
Deciding how long to run an experiment
When you experiment with the available cache settings, you should let the cache
entries fill up to 100% after each adjustment. First, monitor how much time
elapses before the cache is filled with the current settings. Then, when you know
that length of time, run the experiment for twice that long to get meaningful data.
For example, if the cache fills up in 2 days, run your experiment for twice the
amount of time (4 days) to see the effects of your changes on the cache statistics.
Running the experiment for the same amount of time as it takes the cache to fill up
may not provide accurate statistics.
Selecting a cache level
The Instant Internet unit is shipped with three predefined cache levels—
Conservative, Moderate, and Aggressive—and an additional Custom level for
which you can define your own settings. Each cache level sets expiration options
that are applied to entries in the cache that do not have an expiration date and time
stamp.
Note:
Be sure to adjust only one or two settings at a time to make it
easier to measure the results of your changes.
Note:
To ensure that the statistics are accurate with your new settings,
reset the cache statistics before you begin each experiment. For details,
refer to
“Resetting cache statistics” on page 222 .
Note:
The Instant Internet unit always uses the actual expiration date and
time if the originator of a Web entry has set them for an entry.