[rrd-users] Re: Performance (Was: Help with Bash script to calc end-time in multiples of 300 ?)

On Mon, Nov 20, 2006 at 11:21:18PM +0000, Simon Hobson wrote:

> >E.g. 288 pixels, 300 seconds per interval, equals one day (using an RRA
> >of 1 PDP per CDP).
> >
> >336 pixels, 6 * 300 seconds per interval, equals one week (using an RRA
> >of 6 PDPs per CDP).
>
> None of us have screens that small ;-)
>
> Apart from which, the text underneath the graphs already runs out of
> room at 600 pixels wide ! So I've got (IIRC) 25hr x 5min, 12 1/2d x
> 1/2hr, 50d x 2hr, and 300d x 12hr - strange periods indeed, but
> that's what the data happens to come out at.

25 hours in stead of 24 hours. This means the width is 300 pixels where
I used 288 pixels. Given the rest of the example, I guess you may
actually be looking at 50 hours (600 pixels)?

I think you can double (or triple) the amount of pixels without much
speed penalty (but do verify this before you change everything).
Wider graphs do mean more work (and more memory).

Of course, if you prefer to have equal size graphs, this won't do
you any good.

You don't have to display all legend text on one row. I often do this:
min avg max
bytes in: 1.23 5.67k 123.45M
bytes out: 10.12 10.12k 10.12M
errors in: 0.00 0.50 10.12
errors out: 0.00 0.50 10.12
[and so on]

Labels should look like "bytes in: " (all equal size, space filled)
and all GPRINT statements use "%6.2lf%s" (all equal size as well)
The first line is a COMMENT starting with spaces.

HTH
--
Alex van den Bogaerdt
http://www.vandenbogaerdt.nl/rrdtool/

--
Unsubscribe mailto:rrd-users-request@list.ee.ethz.ch?subject=unsubscribe
Help mailto:rrd-users-request@list.ee.ethz.ch?subject=help
Archive http://lists.ee.ethz.ch/rrd-users
WebAdmin http://lists.ee.ethz.ch/lsg2.cgi