Re: back-bdb future

["Kurt D. Zeilenga" <Kurt@OpenLDAP.org>]

At 05:36 AM 3/16/2003, Jonghyuk Choi wrote:




>>For 2.2, we should probably consider using lists of IDL ranges. Could get
>>ugly, but it might be more useful for large databases.
>
>Another option would be to switch to coarse grained IDL which points to
>a fixed range of consecurive IDs. It's effectively the same as changing
>the ID's base to 2^n. Entire IDL or portions of it could be changed to
>use this coarse grained ID. Advantages are 1) simple list management and
>2) bounded overhead. Worst case 2^n times more IDs. Common case overhead
>would be very low, if we assume there's a locality of IDs (ex for dn2id
>subtree).

I note that dn2id is the only index for which we maintain any
locality of reference.