Open LDAP query slow for 5k users only (each having 40 entries)

[Vishal Mathur <vishal@netyantra.com>]

Hi all,

I have some more issues regarding the same problem.
This is very important as for a single call I cann't affored to have 15 
second
just to search the user information.

Issue 1 => do ldap search uses indexs even when filter is not specified ?
say i want to search all entries (filter null) under dn,
 profiletype=userprofile,pid=vishal@ny.com,dc=mydomain
Can it help if i create index on objectclass etc.
When i created index on objectClass my search started taking 25% extra 
time(.2 to .25 seconds.). Then i indexed both pid (mainly used attribute 
in all entry) and objectclass it started taking even more time. 
slapindex should help as i change my slapd.conf with index ?? Why i need 
to use slapdadd as entries are already present.

Issue 2 =>
I have many attributes which can be taken as integers like registration 
time
on server, max simultanous connection reguired.  Currently i have to SUP 
all my attributes with name object.
  attributetype ( 10.1.1 NAME 'regTime' SUP name SINGLE-VALUE )
Can i use any integer object class ?
if yes then how much it will help to have integer instead of 
strings/names ?
How do i get this interger values in my C client ?

Issue 3 => length of dn, attribute and object class name etc.
Does it help ldap to search if my dn length is smaller.
Say if i want a tree level (sub) search on base dn
 profiletype=userprofile,pid=vishal@ny.com,dc=mydomain
There are arround 5 k dn in the same level with same kind of naming 
style. Does it help searching ldap if i rename it to
 ptype=up,p=vishal@ny.com,dc=m

Issue 4 => memory sizes ??
cachesize 10000
dbcachesize 1000000
Are both the sizes in kb ? what is optimal limit ?
which variable i need to tune to say cache all 10 k profiles dn in the 
memory and not the attributes/values ? I can see this memory is not used 
at startup from the ldap. it use as and when needed, in my case 
afterwards it started
swapping on my machine. Can i configure to take all memory at startup ?

Issue 5 => Tree style.
I changed my tree style from

dc=root
   pid=101@netyantra.com,dc=nycube
   pid=102@netyantra.com,dc=nycube
                            ..... 10 k elments in same level.

To new stype

dc=root
  pid=netyantra.com,dc=nycube
         pid=101,pid=netyantra.com,dc=nycube
         pid=101,pid=netyantra.com,dc=nycube
           ---- arround 5 k entry
   pid=example.com,dc=nycube
         pid=101,pid=example.com,dc=nycube
         pid=101,pid=example.com,dc=nycube
           ---- arround 5 k entry

STILL time taken per user search is 15 sec. which is too high.

Regards,

--
Vishal Mathur
Manager - Applications Group
Netyantra India Pvt. Ltd.
3rd Floor, Jaysynth Center, Plot No. 6,
Sector 24, Turbhe, Mumbai - 400 705
Phone: 91 22 27832547/9

-------------------------------------------------------------------------------
Disclaimer -
If you are not the intended recipient of this transmission to whom it is
addressed, or have received this transmission in error, you are hereby
notified that any dissemination, distribution or copying of this transmission
is strictly prohibited. Please notify us immediately and delete this e-mail
from your system. The sender does not accept liability for any errors or
omissions in the contents of this message which arise as a result of e-mail
transmission, which cannot be guaranteed to be secure or error-free as
information could be intercepted, corrupted, lost, destroyed, arrive late or
incomplete, arrive at wrong address or contain viruses. If verification
is required please request a hard-copy version.  This e-mail contains only the
personal opinions of the sender and does not represent an official
communication from NetYantra of any manner.
-------------------------------------------------------------------------------