Clustering the SP

[Cantor, Scott]

On 1/22/14, 10:29 AM, "Martin Haase" <Martin.Haase at DAASI.de> wrote:
>
>thank you for your time. I understand that there's an ODBC resource
>problem, and that the best option is not to use SP session at all once
>the app has its session.

Well, possibly, but using worker mode, which is how *all* Apache servers
in 2014 (or frankly 2005) should be configured, may well work fine.

I know for a fact that PHP can be configured to allow for worker mode by
using an out of process PHP engine with a process pool. So that is the
answer.

>But what about the shared shibd? What I do not understand is that "the
>remoting was not meant
> to work across machines". The Wiki recommends that many machines can
>share a common shibd.

It says they can, and that it may work for low volumes.

> What if I put a load balancer in front of two shibd's and, if the active
>fails, the LB switches over to the passive? I understand all sessions
>from the active shibd node will be lost then, but I could live with that.

I believe based on the mere fact you have some kind of load issue that it
will not scale even close to what's needed. I've seen it kill performance.

-- Scott