Implementing Caching with Standard OpenLDAP

1
Implementing Caching with Standard OpenLDAP

• Johan Jönemo, Lund University

2
GlobusLDAP ? OpenLDAP

• Why Standard OpenLDAP is already widely spread
  and accepted. It is also maintained and updated
  frequently to address e.g. security issues.
• How Different approaches were considered (and
  tried). The plethora of back-ends and overlays
  seemed to offer rich possibilities.
• However A different solution was chosen.

3
How apparently not to do it

• OpenLDAP has several back-ends that tie it to
  different scripting languages as well as
  different overlays to modify its behaviour.
• Unfortunately they lie slightly beside the main
  development main stream. They each seem to be
  either experimental or obsolete.
• Furthermore or quite possibly as a consequence of
  this, in practice, OpenLDAP is often distributed
  with a limited selection of back-ends and
  overlays.

4
Instead do this

• Configure a normal (almost) directory.
• Run the existing information provider code
  periodically and insert the information in the
  directory.
• This gives fast responses to queries, only a
  static directory has to be queried.

5
Details

• Based on back-hdb, a back-end that moves subtrees
  in constant time while being transaction-safe.
  This is in practice always included in every
  OpenLDAP package.
• Builds in a separate build-tree and switches
  trees to keep consistency.

6
Current Status

• Alpha
• Adding features for
• Tweaking update frequency
• Tweaking priorities
• Suggest more by emailing me...