0

Someday ago I'd experimented with different postgres builds. I was changing build configuration parameters to see the impact on queries execution times. My memory can fail me, but I think I did a script that "permuted" many postgres build options from Building and Installation with Autoconf and Make. With the various builds I did run tests that I thought would showcase a change. Schemas were synthetic and some kind of reminiscence of what I had worked on those days. One thing that did not show any difference for my test was BLCKSZ(--with-wal-blocksize=BLOCKSIZE).

Anyone know real life examples of BLCKSZ doing improvement? More specifically, database size, schemas, queries. Anything that I can reproduce, at least synthetically.

Improve this question
13
  • Hi, and welcome to dba.se! You might be interested in this.
    – Vérace
    Commented Dec 5 at 4:47
  • @Vérace hi, the link is irrelevant to the question.
    – user14063792468
    Commented Dec 5 at 11:52
  • @Vérace hi, the link has some insights, yes. but, what concerns me is there are lots of might word involved. i myself tested for the possible. at least i want to see positive answers, like,- yes, we had seen the case.
    – user14063792468
    Commented Dec 5 at 12:25
  • 1
    Well, all I can say is that I've yet to see a case where advice was given that in case X, you change the BLCKSZ to Y - it doesn't appear to be an issue - this is the only time I've seen it proposed, but Craig Ringer - a big hitter in the PostgreSQL sphere - proposed a different solution - and partitioning has come a long way since version 12! Plus, the fact that Erwin Brandstetter has never done it says a lot to me (hint, check his rep. here and on SO).
    – Vérace
    Commented Dec 5 at 19:06
  • @Craig Ringer's comment, below his answer, clarifies things up.
    – user14063792468
    Commented Dec 6 at 1:10

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.