IM (since 5.0) does collect some data even if you weren't charting it; these are called 'auto-record variables'. By default that includes RTT and packet loss. Others can be added per-probe, via an 'autorecord' section in the probe file. There's a little description of how that works in the developer docs, under Creating Custom Probes > Custom Probe Overview > Recording Probe Data. If nothing is charting it, that data is kept around for one day, so you need to be reporting to IMDatabase to take advantage of it.

In terms of long-term trending, every individual datapoint is present in the database, until it gets expired by the device or interface's retention policy. The database also generates daily, hourly and custom (i.e. 5-minute, 30-minute, etc.) samples, in the datasampledaily, datasamplehourly and datasample tables, respectively. Each sample contains the min, max, mean and standard deviation of all the datapoints over that period.

This was a little difficult to use in 5.0, because samples would only get created once the policy's 'raw expiration' had passed. In 5.1, samples are updated in real-time, so you can see the min/max/mean change as new datapoints come in. That should make it easier to do trending from just those tables, without having to look at the datapoint table.


Gordon, Andrey wrote:
1. See long term trending on the interfaces (and other counters)
2. Being able to access any graph without first setting it up. In other words, 
if I have a 3560 with 48 ports, and I did not create a graph of every port, is 
it still possible to retrieve historical data from IMDB? I've noticed that 
sometimes when I create a graph for something, the historical data is already 
there even though i wasnt' graphing that counter before. I just assumed that 
that's coming from IMDB.

--
David Schnur
Dartware, LLC
http://www.dartware.com
____________________________________________________________________
List archives: http://www.mail-archive.com/intermapper-talk%40list.dartware.com/
To unsubscribe: send email to: [email protected]

Reply via email to