2012/8/26 Tom Lane <tgl@sss.pgh.pa.us>:
> Pavel Stehule <pavel.stehule@gmail.com> writes:
>> is possible use seqscan for index?
>
> No, not for a normal indexscan --- concurrent page splits would break
> it.
>
and what about seq scan for prefetch index - and processing should be
random, but over pages in cache?
Pavel
> VACUUM can do that, mainly because it doesn't care if it visits some
> entries twice (and even then, it has to add a lot of pushups to ensure
> it doesn't miss any entries).
>
> regards, tom lane