On 19.9.2005, at 22:27, Shane Dawalt wrote:
Vladimir,
So you basically set up your own database scheme that stores raw
data. Did you then write your own data analysis code to perform
whatever statistical work you needed for billing? Do you provide
usage graphs from this databased information too?
The system I have developed stores raw data computed from flow files in
Berkeley DB. From there they are aggregated (e.g. 5min period is
aggregated to week period) and exported to SQL-based DB. The
statistical work is done in the SQL-based DB. (which is not my
responsibility)
The graphs are provided as a separate branch of data storage - the
script which stores the data into Berkeley DB stores them into RRD file
as well. The RRD file is then used for drawing graphs whreas billing
data are computed from Berkeley DB.
(The whole system is actually more complicated)
My goal was to keep code-writing to a minimum - clearly an
unattainable goal. :-) I expected some coding for the billing
portion, but I fully expected something to exist for storing real
data, generating graphs for the data, and providing a way to get at
the real data for 95th percentile report generation.
Two remarks:
- after completing work on proprietary Netflow-based billing system I
strongly belive that every billing system has to be proprietary :)
- one way to avoid the statistical work is to offload it to the IT
department :)
v.
_______________________________________________
Flow-tools mailing list
[EMAIL PROTECTED]
http://mailman.splintered.net/mailman/listinfo/flow-tools
