Re: [Tracker] using tracker framework for other data

["Adrien Bustany" <abustany gnome org>]

> Hi Ivan,
>
> Thank you for your answer. It sound that introspection in this
> direction is possible.
>
> > ÂI am very curious on how tracker performs on such a big specific
> > dataset. It is not designed for that (our mind is on user data and
> > embedded devices) but can be an excellent test!
> What do you mean about the limitation of the size. In our use case
> each entry have a small size, for example an abstract of one
> publication. However, the whole database should store about ~50Gb of
> entries (for ~15 ontology). Except the case of the sqlite backend,
> does this could be an strong issue?
I don't know of anybody having tried tracker with such a big dataset...
Actually, the optimizations you make for embedded devices and for big
servers are not really the same :p

Another solution would be to adapt libtracker-data to make tracker run on
top of postgres, but that's probably more work and I've no idea how it
would perform. I would suggest converting your dataset to turtle, importing
it in Tracker and testing a few queries... If it's too slow, I have been
using virtuoso in the past and we handled datasets of several hundreds of
GB without too much problems (you still need a decent machine to run it
on, and Tracker is *much* easier to use).

Cheers

Adrien

PS. For big datasets, we had also evaluated Oracle 11g, but we dropped it
for some reason I forgot... Maybe because it was too expensive :p

> > Âgood luck and feel free to post questions or write on IRC!
> Thank you, I will :)
>
> Regards,
> Olivier
>
> --
> -------------------------------
> olivier.gelux.ch
> _______________________________________________
> tracker-list mailing list
> tracker-list gnome org
> http://mail.gnome.org/mailman/listinfo/tracker-list