Comparing Postgres connection pooler support for prepared statements | POSETTE 2024

[@byterelay]

Great summary of what happens in prep stmts. Thanks, Jelte! Of course, we will fix those bugs in Odyssey...

[@HolgerJakobs]

@2:05 It would have been better to show a 2nd client. It's not likely that the same client opens a second connection to the same server (or pooler, for that matter). @5:40 The problem can be mitigated if all names for prepared statements are created including the connection id (pid of the backend). No real need for a pooler to do this, if clients are well-behaved.