Re: [Tracker] Clues regarding improving performance of tracker-store

[Jonatan Pålsson <jonatan palsson pelagicore com>]

On 18 July 2013 17:44, Martyn Russell <martyn lanedo com> wrote:
> On 15/07/13 08:08, Jonatan Pålsson wrote:
> > On 12 July 2013 18:52, Ivan Frade <ivan frade gmail com> wrote:
> > > Hi!
> > >
> > >   If you are not using email, that ontology will be translated into some
> > > tables in the database but they are never touched. It should not make big
> > > difference.
> > > >
> > Aha, this is very good to know. It should mean I can keep several
> > flattened ontologies side-by-side them impacting each other
> > negatively. This is good!
>
>
> Generally speaking, you want to aim for the least changes as possible to
> achieve what you want. That way you keep closer to upstream and have less
> maintenance burden.
>
> What's not clear to me is your targets or if you're looking for
> optimisations out of curiosity.
>
> What do you want to improve, query or insert times? Disk or memory usage,
> etc. etc. The list goes on. It's all possible, but you sacrifice something
> else when you do this.
At this point, the optimizations are mostly out of curiosity, in order
to establish a fastest possible insertion speed. I am looking for a
way to insert the (most likely multimedia) metadata contents of a
(probably) removable device as fast as possible. I realize lookup
speeds may be negatively impacted by this, which is of course not
ideal, but if kept "low" might be acceptable. I would have to
evaluate, and, as you say, play with the sliding scale ;)

Lowering disk usage is also of interest, but I will keep these matters
separate unless one seems to massively impact the other, and look at
insertion/lookup speed first. I still haven't started looking at this
however.

Thanks for the input, and any other suggestions for improving
insertion speed are greatly appreciated!

--
Regards,
Jonatan Pålsson

Pelagicore AB
Ekelundsgatan 4, 6th floor, SE-411 18 Gothenburg, Sweden