[rtg] rtg status?

[Charles Gucker]

On Wed, Jan 7, 2009 at 8:18 PM, Jacob Steinberger
<trefalgar at realitybytes.net> wrote:
> Matt Provost wrote:
>>
>> There's a lot that I'd like to do but we'll see how much time I have.
>> The problem with the vast amounts of data that rtg collects is an
>> interesting one. There's lots of good research out there about
>> compressing time series data, but I'm not sure how you would do it and
>> still keep it in SQL. Then again, you could write a backend for the
>> poller that writes to any format...
>>
>
> In all honesty, if I wanted to compress/average data based on time, I'd use
> another product. I use RTG for two reasons - MySQL, and raw data. I'd hate
> to see either go away, in any iteration.

Well, what we did was create another DB on our webserver (which does
the front-end for customers) and do remote queries to the RTG DB.
via cron, we aggregate "cache" data for generating less granular
graphs, but if a user requests a specific time, which requires more
granular data (billing info included) we go back to the RTG DB
directly.    It really speeds up the usage graphing routines and
greatly reduces the hits on the RTG DB since most hits are kept on the
front-end server.    We haven't done it yet, but we can also archive
the older granular data, but still be able to generate weekly, monthly
or yearly less granular graphs (much like RRD graphs).

But we would need to do a lot of work to the code to make it release
able, but certainly something that can be done.

charles