[Nepomuk] Enterprise Nepomuk proof of concept?

[Bryce L Nordgren]

Hi list,

I have a question about how amenable Nepomuk is to scaling up and out.
Specifically, if I wanted to create "a *physically and semantically
decentralized* document management system, relevant to a large, distributed
research organization, capable of handling Big Data", are we talking
months, years, or decades?

Essentially, I'm talking about RDF annotations concerning the contents of
shared disks within a large organization. There should be very little
indexing taking place on desktops, although desktops should be able to
query an index generated by a "Nepomuk cluster" in the data center.
Annotations by end users, of course, should be aggregated into the central
metadata store. And of course, users should only see results that file
permissions allow them to see.

The gist of the idea is described here:
https://collab.firelab.org/software/projects/semantic-tech/wiki/Document_Management#Form-of-the-system

I understand that certain components (notably Virtuoso) are cluster
friendly. My concern lies primarily with the reliance on dbus, which is
bounded by a single physical workstation, if I understand correctly. Is
this less of an obstacle than I am fearing (e.g., is it possible to make a
"dbus adapter" which relays updates and queries off to a separate store)?

For the moment, I'm just talking about a proof of concept system, not a
production-ready one.

Thank you for your time,
Bryce
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/nepomuk/attachments/20120204/7bf72286/attachment.html>