"Mitch Vincent" <mitch@huntsvilleal.com> writes:
>> So, may I have the envelope please? What's the timings?
> 6.5.3:
> ! system usage stats:
> ! 0.436792 elapsed 0.275139 user 0.157033 system sec
> ! [0.283135 user 0.173026 sys total]
> ! 0/0 [0/0] filesystem blocks in/out
> ! 0/149 [0/332] page faults/reclaims, 0 [0] swaps
> ! 0 [0] signals rcvd, 0/0 [2/2] messages rcvd/sent
> ! 0/8 [2/9] voluntary/involuntary context switches
> ! postgres usage stats:
> ! Shared blocks: 1403 read, 0 written, buffer hit rate
> = 51.22%
> ! Local blocks: 0 read, 0 written, buffer hit rate
> = 0.00%
> ! Direct blocks: 0 read, 0 written
> 7.0 :
> ! system usage stats:
> ! 1.461997 elapsed 1.224377 user 0.234618 system sec
> ! [1.238219 user 0.255382 sys total]
> ! 0/12 [0/12] filesystem blocks in/out
> ! 0/60 [0/318] page faults/reclaims, 0 [0] swaps
> ! 0 [0] signals rcvd, 0/0 [2/2] messages rcvd/sent
> ! 0/22 [1/24] voluntary/involuntary context switches
> ! postgres usage stats:
> ! Shared blocks: 2713 read, 0 written, buffer hit rate
> = 25.34%
> ! Local blocks: 0 read, 0 written, buffer hit rate
> = 0.00%
> ! Direct blocks: 0 read, 0 written
Well, drat. Looks like 7.0's query plan is slower :-(. There's
something fishy about the numbers for 6.5.3 though --- how could it have
done that query with zero blocks read? Are you sure you are comparing
apples to apples here? I wonder whether the 6.5 system already had the
tables cached in kernel disk buffers while 7.0 was working from a
standing start and had to physically go to the disk. Also, did both
versions have the same -B and -S settings?
regards, tom lane
