The following are notes from an Adobe MAX 2007 session.

Server monitoring can help you get rid of some of your bottlenecks. Can make a big difference. The general server monitoring can even be used in production,  but DO NOT use the memory monitoring on a production site. May bring it down, even on low load.

The Overview Screen will let you see slow active requests, charts of the number of requests, average response time, etc. What errors have been thrown lately? What alerts have been posted (will explain later)? Various other statistics like that.

The Statistics Screen will display a bunch of various useful bits of info.

Requests. Active CF threads, slowest threads, active sessions, cumulative server usage, highest hit counts, cache status, etc. By looking at this info, you can find long-running active requests that can actually wreak havoc on the whole server. You start chopping those down, and before you know it, your server will run like a race horse. There’s also a memory statistics section.

Memory. Shows general memory usage, requests by memory usage, threads by memory usage, queries by memory usage, sessions by memory usage, application scope mrmory usage. You may at times find large amounts of memory that aren’t even used effectively in your apps, thus can slightly tweak the way things are being done. Beware: Garbage Collection isn’t always the answer. Memory can be hidden in shared scopes or caches.

Database. Active queries, slowest queries, cached queries, etc. Most frequently run queries.

Errors. Requests with errors, requests that have timed out. Having a lot of errors used to cause stability issues; not as much now, although if your logs fill up the whole hard drive, ColdFusion will have a problem! Plus, if you’re having that many errors, won’t you want to know about it?

Alerts. Can set alerts for various circumstances you’ll want to be notified on. If the hung thread count reaches a certain maximum, or if a thread has run for x seconds, you can do certain actions (send email, dump snapshot, kill thread, reject any more requests) or you can supply a processing CFC that can then do anything you want.

The Snapshots Screen will give you all the details of the server at the moment the snapshot was taken. Will save all the detail into a text file. Can’t automate regular snapshots, although you probably could do it with the Admin API. Just set up a scheduled task that tells the Admin API to take a snapshot.