Re: repeated attribute search...

[Matthew Backes <lucca@csun.edu>]

On Wednesday, May 7, 2003, at 13:36 US/Pacific, Raj Balasubramanian 
wrote:

> Hello All
> I was wondering if you saw any major performance degrade when a search 
> is performed for an attribute thats referenced multiple times within a 
> search string. For example: 
> (myid=0909)(CN=raj)(myid=0909)(uid=189)(myid=0909)

I had some buggy code that generated filters like this for a while and 
I noticed no significant performance impact.

I've also made extensive use of large filters (where the filter text is 
say, 40k or the like) and I noticed no significant performance impacts 
from that easier.  There are some cases where pathologically large 
filters are not considered valid.  I haven't tracked this down as it 
isn't something I generally need; there are much better ways to get 
that sort of data out of a directory server.  I suspect that is an 
issue with protocol limits to the filter size.

It is important to note that if it searches for (myid=0909) multiple 
times separately, especially in separate connections, performance will 
suffer simply because of the overhead of starting Another search or 
Another connection.  Even more so if these are separate SSL/TLS 
connections.

Matthew Backes
lucca@csun.edu