This post originated from an RSS feed registered with Web Buzz
by Mark Masterson.
Original Post: Logging 2.0
Feed Title: Process Perfection
Feed URL: http://www.jroller.com/MasterMark/feed/entries/rss
Feed Description: Life, the Universe, workflow, BPM, Java, Ruby, functional/generative/meta programming, pi calculus, patterns, the Grid, agents, software architecture and the kitchen sink. :)
One of the reasons Tim Bray's WF thing got my attention was that log management is a topic near and dear to my heart. Kind of embarrassing to admit - logs aren't exactly sexy. In a prior life, I wrote the appender for log4J that transformed logging events into SNMP traps. That code was later integrated into both JBoss and AdventNet's Java SDK, and used in lots of enterprisey contexts. I myself have used it at a mega bank and a mega insurer, and it did its job well enough.
So I think I'm allowed to assert that I know something about logging. Red-headed stepchild it is, but the fact is, most enterprises would grind to a halt overnight without it. So giving the topic some attention is surely a good idea.
Certainly everything Kevin says in his post is spot on. In particular, this:
The current feedback loops providing by log data tends to be glacial
... hit a nerve. That was the problem that the SNMPTrapAppender was designed to solve. Mountains of log files, which nobody ever even glances at, until after there's a problem. The idea behind my appender was to send the important events to a management console, where they would get noticed, and acted on immediately. Feedback loop problem = solved. Right? Yeah, right. In the real world, it just shifted the problem. The console operators paid no attention to the messages, applications either logged too much or too little, etc. Appenders like mine do help, sure, but... There's no silver bullet.
I'm like a broken record scratched CD, playing a woeful Fred Brooks tune... Must stop.
Previous Topic
