I just watched the multi-tenancy webinar http://www.vimeo.com/20296211 and had a few questions about what was discussed.

1. It was stated that if you're using a separate schema approach you need to turn off the lvl 2 cache. The idea was that customer 1 would be present in all schema's. If you implement a id generator such that the id's were unique across the entire set of schemas would it be safe to use a lvl 2 cache?

2. At one point Steve said Hibernate 4 was going to handle a lot of this but that it was still in discussion. Is there any documentation that shows what was decided and how it was implemented?

Well specifically I believe I said that without the added support *in* Hibernate you would need to turn off second level cache. And yes, even without the multi-tenancy support added in Hibernate 4 if you could implement an id generator that generated unique ids across all tenants (uuid, for example) then caching could be used.


Not sure to what you are referring specifically. Hibernate 4 already has multi-tenancy support for the separate schema approaches. What is lacking is support for discriminated approaches. That will come in 4.1 or later.