On 06/01/2014 22:03, Michael Ströder wrote:
(I take this point [email protected] since it discusses
OpenLDAP-specific things.)
Howard Chu wrote:
The discussion of caching here
http://www.ietf.org/id/draft-bannister-dbis-mapping-02.txt is one such example
- this is purely a client-side implementation issue. Also you give nscd as an
example, and nscd has been thoroughly discredited and is well known to be
unsuitable for real use. Critical deployments can use a local LDAP server with
a replica of the central data, to avoid error-prone caching implementations.
This is a commonly recommended approach when using OpenLDAP nssov, for example.
I really wonder how this replication approach works in practice without
disclosing too much data on a system more exposed to attacks from the outside.
In theory one could implement partial replication based on systems's bind
identity. But in practice I have some doubts because in a really paranoid
setup you don't even want to disclose replication meta data and intermediate
entries of the tree structure.
Ciao, Michael.
I would prefer to have the option of using a caching implementation that
was not error-prone, rather than a local LDAP server with replication.
This is because setting up local LDAP servers with replication are not
necessarily straightforward, deciding where to implement them becomes
troublesome in a large estate, where they are not implemented you may
very get poor performance, and not all LDAP servers are supported in a
local deployment. Not to mention the security implications that Michael
mentions above.
Best regards,
Mark.
