(ITS#7514) Optimize OpenLDAP with paged results

[quanah@OpenLDAP.org]

Full_Name: Quanah Gibson-Mount
Version: 2.4.33
OS: Linux 2.6
URL: ftp://ftp.openldap.org/incoming/
Submission from: (NULL) (74.196.25.250)


Currently, searches of back-mdb using paged results are extremely slow.  Up to
300 times slower than when not using paged results:

Without paged results:

 time  ldapsearch -LLL -x -H ldap://zre-ldap002.eng.vmware.com -b
"ou=people,dc=xxxx,dc=com" -s sub -D uid=zimbra,cn=admins,cn=zimbra -w zimbra
'(objectClass=zimbraAccount)' uid

real    0m0.004s
user    0m0.010s
sys     0m0.000s

With paged results:

zimbra@zre-ldap002:~$ time  ldapsearch -LLL -x -H
ldap://zre-ldap002.eng.vmware.com -b "ou=people,dc=xxxxx,dc=com" -s sub -D
uid=zimbra,cn=admins,cn=zimbra -w zimbra '(objectClass=zimbraAccount)' uid -E
pr=1000

real    0m1.204s
user    0m0.010s
sys     0m0.000s


Using paged results is fairly standard in Java clients, to prevent them from
hitting OOME's based on the way in which results are returned into a single
object.