On Fri, Aug 5, 2016 at 12:25:39PM -0300, Claudio Freire wrote:
> > 2. The existence of index pointers to intermediate tuples will lead to index
> > bloat. This is the same problem that we found in Simon's original proposal
> > (discussed internally). None of the intermediate line pointers can be
> > vacuumed until the entire chain becomes DEAD. Event if the a duplicate index
> > key is inserted for every index, knowing that and reclaiming to the index
> > pointers to the original root line pointer, will be difficult.
>
> I don't see the difference it makes for bloat between storing the root
> tid and the intermediate tid, but I haven't yet figured out how
> vacuuming would work so maybe I have to think about that to realize
> the difference.
Think of page pruning --- we can't remove a ctid that an index points
to. The more ctids you point to in a HOT chain, the fewer ctids you can
remove --- that's why we want to point only to the head of the HOT/WARM
chain.
-- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB
http://enterprisedb.com
+ As you are, so once was I. As I am, so you will be. +
+ Ancient Roman grave inscription +
