Friday, 23 April 2010

Sprint 2 plan

Following the Kick-off meeting with Epimorphics, I've updated the first sprint plan (http://code.google.com/p/milarq/wiki/SprintPlan_1) and prepared the second one (http://code.google.com/p/milarq/wiki/SprintPlan_2).

There was no formal review of the first sprint, but there wasn't really much to review. And the Epimorphics meeting really was the planning session for the second sprint. The next such meeting will probably combine sprint review and planning.

The main work of sprint 2 will be to perform some scoping experiments for SPARQL queries against the CLAROS database, to get some early indications of what kinds of performance improvement might be expected through different query handling strategies.

Monday, 19 April 2010

Kick-off meeting with Epimorphics

A two-day project kick-off meeting with Epimorphics was help last week (15-16 April), which I think went very well.

All intended management and technical goals for the meeting were achieved, with goals confirmed, roles clarified, a technical plan in place and a framework for evaluation in place.

The main project success criterion is that we can improve query performance sufficiently to take CLAROS public.  Roughly, this means achieving sub-second response times for all queries.

During the project, Epimorphics will undertake exploration and modification of the query mechanisms, while I will suppiort the test and evaluation elements by applying any agreed query redesigns to the test suite, and evaluate revised query handling mechanisms in the full CLAROS system.

In the first month, Epimorhics will perform a series of scoping experiments, to explore the unknown elements in the project plan, and "de-risk" the project.  After this,we will hold a second meeting to decide concrete steps for further development, by which time the actual steps that need to be performed should be much clearer.

During the two days, we installed the CLAROS query software and test suite from its various subversion repositories and loaded up a copy of the data currently running on the CLAROS demonstration server at OeRC.  The resulting system passed all test cases in the test suite, which represent instances of all the queries used by the CLAROS Explorer VRE application.

All-in-all, I felt this was a positive and successful meeting.