[rrd-users] RRA/RRD tuning

[Christoph Anton Mitterer]

On Mon, 2012-08-13 at 15:24 +0200, Tobias Oetiker wrote:
> you have to work on the english though ... tuneing may not be the
> word you are looking for in your context ... 
I meant tuning obviously, which doesn't fit too bad IMHO (?!),... OTOH,
I'm fine with optimisation, too.

> * optimizing update time
> * optimizing storage space
> * optimizing graphing time
Uhm... wasn't that exactly what I did?
A first section, with general information about RRAs
 1) Time Spans
 3) Granularity
A second section about optimisation with:
 1) Tuning of the size of a RRD
 2) Tuning of the time needed for updates
 3) Tuning of the time needed to retrieve and process data, mainly with
respect to graphing

So please tell me more clearly, what you expect :)


> > - We might add which CF is faster...
> > I mean MAX/MIN are likely just comparisons and are therefore a tiny bit
> > faster (CPU wise) as e.g. AVERAGE (which I guess includes some adding
> > and dividing?).
> I don't think that this will make much of a difference compared to
> disk I/O
Yeah,... likely,... which is also the reason why I haven't had included
it yet.


> You do not realy elaborate much about how to choose the RRA sizes
> for optimizing reading operations ...
Uhm... what/how exactly do you mean?

> as I said ... the language needs work
Phew... well... guess that's my English (apart from spelling mistakes,
obviously; I have a new notebook since a few days and need to get used
to it's keyboard ^^)... and as I work here in one of Germany's biggest
research centres with predominantly English speakers (including native
ones), who seem to understand me quite well, you need to point me more
directly on what you don't like =)


> I guess a NOTE section in the rrd create document might be helpful
> 
> but if you expand this further, you may also create your own page
> and add a link to it to the rrd create page ...
Well,... I personally haven't intended to expand this much more, mainly
because I don't know much more on how to tune RRDs/RRAs by themselves...
and I guess the general topics of optimising IO with primary
random-access goes beyond the scope of rrdtool (or what it's
documentation should deliver).
I once made a small research lecture about all aspects of block layer
alignment, which I could perhaps link to.

But... I think that placing that information in rrdcreate "hides" it a
bit from the user, so maybe a general rrd-optimisations document where
all people can add reasonable stuff would be a good idea.


Cheers,
Chris.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 5502 bytes
Desc: not available
Url : http://lists.oetiker.ch/pipermail/rrd-users/attachments/20120813/b95244ee/attachment.bin