At 09:39 AM 5/06/2011, heineferreira wrote:

>Hi
>
>How much does the memory usage differ between classic and super classic?

Not much. Both models use separate DB caches per connection. There is a slight win for SC under some conditions, due to in-memory lock structures sharing space.

>Yes I know the difference in architectures - I saw the diagram in the FAQs. It looks like super classic is a better option for smp.

Why do you think that?

>Super Classic was only recently released, so I am concerned about it's stability?

Its stability will be the same as Classic. It's the same engine, deployed with its own listener daemon and some shared memory benefits, instead of using xinetd and running each connection in its own self-contained resource set.

>Also, can you develop on super server and then deploy to super classic?

Yes. Just make certain you reconfigure the DB cache appropriately for the SC deployment.

>Any limitations about classic or super classic that I should know about?

Don't try to use it on a 32-bit OS under production conditions. The idea of SC is to improve the engine's ability to use a well-resourced host machine. It cannot bypass the per-process memory limitations inherent in a 32-bit OS. The resources available to a 32-bit OS will soon be exhausted with multiple users, even if none of them is doing very much.

./heLen