1-/2-dimensional indexes for common columns, rationale? - Mailing list pgsql-performance

Hello,

we have several indexes such as:

create index foo1 on bla (a);
create index foo2 on bla (b);
create index foo3 on bla (a,b);

They are all used often by frequently used queries (according to
pg_statio_user_indexes), but we need somewhat higher INSERT/UPDATE
performance (having tuned most other things) so we'd like to remove some.

Which of the above would generally speaking be most redundant / best to
remove? Is a 2-dimensional index always much slower than a 1-dimensional
with the first column for queries on the first column? Any other
suggestions?

Thanks,
  Marinos