You can benchmark your scenario with and without constraint using a tool like nancy:
https://gitlab.com/postgres-ai/nancy it lets you A/B test different configurations with your own scenarios or using pgbench synthetic workloads. -Michel On Wed, Jul 22, 2020 at 9:27 AM Samuel Nelson <valczir.darkv...@gmail.com> wrote: > I checked, and changing the `bazinga_foo_bar` constraint to: > > alter table bazinga add constraint bazinga_foo_bar foreign key (foo_id, > bar_id) references foo_bar (foo_id, bar_id) deferrable initially deferred; > > seems to fix it to work as we were expecting. Is that particularly > costly? Should I only set the constraint to be deferred when we really > need it? Would it be more efficient to perform the deletes explicitly > within a transaction rather than relying on the cascades and deferring that > one constraint? > > Our resident ex-Oracle DBA said that deferred constraints used to be > heavily recommended against, but he also admitted that he hasn't kept up > with that in the past 10 years. > > -Sam > > https://github.com/nelsam > > "As an adolescent I aspired to lasting fame, I craved factual certainty, > and > I thirsted for a meaningful vision of human life -- so I became a > scientist. > This is like becoming an archbishop so you can meet girls." > -- Matt Cartmill > > > On Wed, Jul 22, 2020 at 10:31 AM David G. Johnston < > david.g.johns...@gmail.com> wrote: > >> On Wed, Jul 22, 2020 at 8:24 AM Samuel Nelson <valczir.darkv...@gmail.com> >> wrote: >> >>> Is there a way to force the delete to cascade to tables in a specific >>> order? >>> >> >> No really, but you can defer constraint checking. >> >> https://www.postgresql.org/docs/12/sql-set-constraints.html >> >> David J. >> >> >
