Sunday, 11 May 2008

Which SIF?

Recently got to know a little more about Capita's Partnership Xchange. OK so it's using the Schools Interoperability Framework transport method and their own SIF based data model to allow SIMS systems to pass core pupil, timetable, attendance and assessment data.

My understanding on the key to it working is the method by which schools determine which data from SIMS is sent out. This is achieved by having particular timetable items being the SIF shared data. Only data associated with pupils in those timetabled sessions is shared. Their core data appears as read only to non-home school SIMS systems. Schools hosting pupils from other schools can however write back attendance and assessment data. this depends on them having Lesson Monitor and Assessment Manager and having mark sheets and timetabled sessions that are the agreed across the consortium.

I think I understand much better now a major issue with SIF. Regardless of the relatively simple data model agreement databases management systems in the Zone have to:
  • Have interfaces to the agreed data model that validate data to the same standards as the most rigorous database receiving the data.
  • Have a method of allocating only that data you want to share
  • Ensure structures of data tables take into account historical data storage, date stamping data etc. so data is not lost
  • Have logic to sort out what to do when data just doesn't match for example timetabled sessions. Just how do you squeeze 5 session attendance data into 6 sessions attendance systems? Nightmare. Data will have to be lost.
Perhaps some of this is simply because we don't have agreed MIS system standards. If for example all School and College MIS systems had to use time to store attendance rather than periods then they could all be matched up. Of course they don't. Getting agreement on architecture will take decades.

1 comment:

Jason Wrage said...

"Just how do you squeeze 5 session attendance data into 6 sessions attendance systems?"
This is a great point, and one of the biggest challenges in implementing SIF from a software/database perspective: getting the number of data entity instances to align from the source system through to all of the potential target systems. Systems that leverage XML internally have an advantage, as do systems that have relatively normalized internal databases. I'm not sure that I would call this issue a nightmare, but it does require vendors to agree on what is reasonable. Unfortunately, the SIF specification itself provides little guidance with respect to this topic.