# [rrd-developers] Using a logarithmic scale on the x-axis?

Philipp Marek philipp at marek.priv.at
Fri Oct 2 10:50:54 CEST 2009

```Hello Sebastian,

>> Now I wondered whether it's possible to get a logarithmic time scale - from "-1
>> month" on the left side, over "-2 weeks", "-1 week", "-3 days", "-1 day", "-12
>> hours", "-6 hours", "-3 hours", "-1 hour", "-30 min" and "-15 min" to "-5 min"
>> as labels.
>
> I guess, this is meant to be an example only, right? Hard-coding those
> values does not make any sense, imho.
Of course; I just wanted to illustrate an example, with steps (factors) of 2 and 3.
(So nice that 24 has so many factors ;-))

> If this is going to be
> implemented, I think it would make sense to calculate a "reasonable"
> value (whatever that might be) for the largest step size from the
> specified time span and determine the smallest step size from the
> highest resolution RRA available for the end of the specified time span.
Well, of course I'd like to specify that the graph should start -1week, -1month or -1day
before.
That the lines (or areas) might be a bit ragged in certain time intervals (when one
higher-resolution store has just been left) is inescapable.

> Calculating reasonable positions for tick marks (and labels) will
> probably be quite tricky.
Well, I'd expect that to be not that much of a problem ...
Try to put the "biggest" markers in there (month, week, day, hour), and then use
multiples of them (with factors of 2 or 3, as appropriate).

>> I think that this could possibly provide a good overview, with more details for
>> the just passed timespans.
>
> Hrm â€¦ I cannot image that there are lots of people out there that are
> able to make a lot of sense from a logarithmic time scale - at least,
> I've never seen something like that before.
Well, I used that quite some time ago, for temperature data ... while the fluctuations
of the last time was clearly visible, the time longer back degenerated to a kind of
trendline ... I found that nice.

And if it's just another parameter, it doesn't have to be turned on, right?

> Anyway, this might be fun just for the sake of it even though (or maybe
> just because) it's non-trivial ;-)
>
>> Is that already implemented (but not documented), or could this be patched in?
>
> So far, something like that has not been implemented.
Well, I'll look from time to time ...
Maybe, if you're the lucky one to implement that, you could tell me when it's been done.

Regards,

Phil

--