As for this PS5 issue we're witnessing, it seems (but it's not an official information though) that when you reach a certain amount of records within SQL Express, you get somewhat an internal loop that increases timeouts. This is what i've been told by IT department in a major worl wide costumer. Nothing granted here, so it's just speculation. On the other hand it could be a handicap on PS5 that, hopefully, would be resolved in PS2009. But as i'm about to migrate both PS5 to PS2009 (hopefully) and from SQL Express to SQL Enterprise, i won't have any clear responses on this. I'll try to search any responses on the web for this issue.
As for the mirroring, unlike clustering where you have failover at the server level, mirroring it's a failover at the databases level only. It's a synchronized replication between 2 servers. One goes down, the mirror takes over. The diddiculty here is that the name of the server changes, and PS doesn't know this server, so...it stops responding. (i got the chance to try this already)
As far as i could perceive, we would have to have an aoutomatic change on the software side (ADMS).
I believe that mirroring is gaining terrain due to the fact that you won't need a shared storage, so you have redundancy there, but it is not as transparent as clustering on the other hand. To try to perceive how fast mirroring is growing and if will costumers start adopting this technology seems an interesting issue to be discussed among you guys. I find that PS is not yet prepared to respond to high availability in a modular way.
With best regards,
David