This package analyzed logs quite well. This is an early version of Traffic Analyzer (1.01), so I feel that if the following few changes were made, the package would be stunning.

Charts and graphs in the "Graph view" should be exportable. Now, charts can only be seen using the Traffic Analyzer software. If these charts were exported as HTML pages, they could be posted online. Now only summary text can be exported

The actual time found in the log record should be used. It appears that Traffic Analyzer uses the time zone offset value in the log record to compute the Greenwich Mean Time (GMT) of each event. This is handy if you are in England, but I would rather have my charts using the local time of the server.

Ability to combine subdomains should be provided. Many ISPs create a separate subdomain for each of their connections and Traffic Analyzer shows them separately. While this is useful for some people, I would rather have summarization at the actual domain level, so that all these people who dial in from the same ISP are summarized together. (See the aol.com and the mediaone.net examples that are not combined.)

Individual data read error statistics should be given. If a log record was bad, it should be in the summary. Bad records could be the result of a server problem or a partially corrupted file, but unless you are able to verify the record count with another program, you can't tell if you have a log format problem or malformed records. Currently you are warned only if ALL the records are unreadable.

Graph detail showing repeated values

Chart width should change for small ranges. At least 10 or so days of data needs to be in your log or the automated charts display poorly. Because the Traffic Analyzer charts are always a constant width, if there is not enough data, data items will be repeated on the legend to fill the width.

Standard deviation or variance values for the averages should be provided. It is nice to see if you are getting consistent traffic or extreme changes in traffic from day to day, which could result in the same average.

Conclusion

Introduction

Copyright © 2001 Rick Smith All rights reserved.