[Building Sakai] Course Management : CmSynchronizer JOB : File Size

[Casey Dunn]

On Fri, May 15, 2009 at 2:43 PM, Thomas Amsler <tpamsler at ucdavis.edu> wrote:

> > Thomas, those are a bit out of date. Are you more interested in how the
> XML
> > is created in the first place?
>
> We are generally interested in how institutions populate the Sakai
> internal CM tables.
> - What modifications are mode to the synchronizer sample code?


a fresh start based on the insight to the CM API provided by the example


>
> - What modifications are made to the CM Hibernate Impl?


additional local columns


>
> - How are the Sakai internal CM tables populated with the initial
> large data set?


the resulting code framework has a number of ways of selecting 'sync
candidates' from
the feed. The original load was done by taking a couple of previous
terms/subject intersections
 at a time and batching them through a 'back office' Sakai / Coursework
instance.

with smaller sets from previous terms establishing performance baselines I
planned out
a back-filling set of jobs and let them run in the back office instance. I
may of used a couple
of dev instances reconfigured to talk to the same DB. having async updates
is fine.



>
> - How frequent CM tables are synchronized with SIS data and how long that
> takes?


originally we would run term or term/subject jobs based on calculated
offering count size.

currently Stanford is running this almost on demand. with any luck they'll
chime in with the
current settings for the event & batch based system.



>
> - How well does the XML sync approach work?


it's fine. the business cycle is now faster than it was with previous
versions of the offering, and
as it can be driven 'on demand' as well support issues are handily resolved.



>
> - How well does the custom scripting sync approach work?


er?


> - Has anybody attempted to load data from SIS-DB to Sakai-CM-Tables
> directly in SQL?
>

Stanford does not allow access to the upstream data w/out going through the
mediating
XML formats.




>
> We have experimented with XML as well as custom groovy scripts that
> use the CmAdmin API.
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://collab.sakaiproject.org/pipermail/sakai-dev/attachments/20090515/4d9618a8/attachment.html