One of the great future innovations that is coming down the pike for administrator types like me is the day when full logging systems become cost effective to run. Right now this isn't normally done except when something happens outside of normal events. The reason is simple, it's cost prohibitive in the three areas of processor time necessary, space used for the logs, and analysis tools/time necessary to wade through the resulting data.
All of this is changing:
Processor time shows few signs of slacking off from its blistering progress pace. It's gotten so bad that processor makers like Intel are starting to seriously branch out into creating new applications that will justify buying their new processors.
Space prices are collapsing as well with both magnetic and optical media dropping pretty quickly.
Finally, the business intelligence tool makers are getting better at creating tools that are simple to use and solve the problem of drilling down quickly and getting useful results out of massive data sets.
At some point, somebody with the appropriate programming skills but without the requisite budget will want business intelligence tools enough to either write it himself or hire some inexpensive programmers to write the tools for him. He won't care for the tools as a money maker in and of themselves. Viewing them as a cost center, he'll open source it to reduce his maintenance costs and we're off to the races.
Posted by TMLutas at February 16, 2004 09:09 AM