> OK, That could even be implemented in ORM proper then. But I ahve to admit, I'd like to see some real-life-ish scenarios where that would be beneficial.

Me too. It's being presented as a great optimisation opportunity in the cloud-tm architecture document.

> So you mean a look up by id where the entity has been deleted already, correct? I am not sure how lazy properties help here.
I can see how it could help if the backend store each property in an individual key like one of the cloud-tm approach but OGM does not do that and I don't think there is much value in this since we use AtomicMap and hopefully, Infinispan will know how to retrieve partial AtomicMap at some point in the future.

I was referring to the individual properties: agree with you that Infinispan should be able to deal with it internally, but I'm pretty sure it doesn't.

We could repurpose this issue to verify my assumptions, then resolve by opening appropriate issues on Infinispan.