Re: Checkpoint cost, looks like it is WAL/CRC

>> Only workable solution would imho be to write the LSN to each 512
byte
>> block (not that I am propagating that idea).

"Only workable" was a stupid formulation, I meant a solution that works
with
a LSN.

> We're not doing anything like that, as it would create an
> impossible space-management problem (or are you happy with
> limiting tuples to 500 bytes?).

To do it, a layer between physical storage and row workmemory
would need to be inserted, of course that would add a lot of overhead.
I guess more overhead than computing a page crc.

> We still don't know enough about the situation to know what a solution
might look like.
> Is the slowdown Josh is seeing due to the extra CPU cost of the CRCs,
or the extra I/O cost,
> or excessive locking of the WAL-related data structures while we do
this stuff, or ???.
> Need more data.

Yes, especially the 10 sec instead of 1 sec response times look very
suspicious.

Andreas