Hi Martin,
> about the scalability issues faced by Simile, covering things like the
> demands that SImile will place on an infrastructure, and looking into issues
> such as whether faceted browsing actually works when faced with very large
> datasets.
As you gathered, the triple store report identifies the current state of
affairs for a survey of triple stores and their handling of a moderately
sized dataset, with some hope that those results might be fairly
extrapolated to extremely large datasets.
The wider picture of what types of scalability issues the project is
facing is quite pertinent to what problems we try to solve and how we go
about solving them, but I don't see us writing a report, per se, on the
topic. Instead, I'll try to identify those issues we're working on
here, with help from the rest of the team as they see fit.
- Store performance
Our system must perform well for users on extremely large datasets (200M
triples?) doing normal read (and in the future insert / update / delete)
operations. This is what the report starts to address at the read level.
- System performance
The system overall, built on the store, will need to perform brilliantly
for users by taking advantage of some precomputation and caching strategies.
- Inferencing
Our system must be able to do inferencing in some bounded time. While
not necessarily a user interaction, it must at least be possible with
extremely large datasets.
- End-user usability (display)
You specifically mentioned the scalability of faceted browsing for users
on large datasets, to which we can say definitely that the answer is no,
Longwell won't scale as-is. How we display volumes of data in such a
way that users can comprehend and navigate them sensibly is another
issue of scale we'll need to address.
- Number of ontologies
The more ontologies we have, the more potential variation there will be
in displaying human-friendly information. There may be some issues in
doing manual mappings between ontologies.
--
Ryan Lee ryanlee_at_w3.org
W3C Research Engineer +1.617.253.5327
http://simile.mit.edu/
Received on Thu Jul 22 2004 - 18:28:56 EDT