Subject Re: [Firebird-Architect] Superserver
Author Dmitry Yemanov
Erik LaBianca wrote:
> What I'd really like to know is the following
> 1. Am I oversimplifying the problem of making classic cluster aware? If
> not it's conceivably something I could see devoting some resources to
> given an agreed upon design.

You're generally correct, but it's not as trivial as it may seem.
Anyway, the appropriate R&D is already in progress.

> 2. Why are people so jazzed about superserver? As I understand it it was
> largely implemented to work around weak ipc on windows. Why push
> superserver on unix then, since classic embedded mode works great, and
> classic has some great potential scalability? Is the lack of a shared
> page cache really that big of a performance hit in the days of 64 bit
> machines with gobs of ram?

With a big per-process cache, the lock manager becomes a bottleneck. So
most of the CS installations use 200-2000 pages for a buffer cache, even
if they have many gigs of RAM. This means that the page cache is not
used very efficiently.

Also, when we speak about more than 1000 connections, a proper
multi-threading outperforms the process based model (less system
resources are required). On Windows, problems arrive with about 600
classic processes, AFAIK.