[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: (ITS#3675) On Directory Server Performance Benchmarking

Hi Howard,

You are right. Actually, this work was done in 2002 when when 2.0.23 was 
just released. At that time, I thought there was no right forum to 
discuss these results. Only recently I realized that the results can be 
made public and the report made available. I guess, I missed the bus :)

I have not really worked on the benchmarking after this project, although 
it might be interesting to apply the same set of experiments to the 
current releases.


On Thu, 21 Apr 2005, Howard Chu wrote:

> I'm curious why you chose such an old release for this study. OpenLDAP 2.0.23 
> was released in February 2002, with the final 2.0 release (2.0.27) in 
> September 2002. There have been two whole release streams since then. Even 
> the 2.1 release is now considered "Historic." As such, any lessons one might 
> have gleaned from studying 2.0's performance characteristics were identified 
> and addressed years ago. For example, we have benchmarks showing that 
> OpenLDAP 2.1 is fully 200 times faster than 2.0, with 2.2 30-50% faster even 
> than 2.1.
> kulkarni@cs.ucr.edu wrote:
>> Full_Name: Dhananjay Kulkarni
>> Version: 2.0.23
>> OS: Mandrake Linux 7.0
>> URL: http://www.cs.ucr.edu/~kulkarni/Dhananjay-Kulkarni-050421.pdf
>> Submission from: (NULL) (
>> LDAP-based directory servers are being widely used in grid computing,
>> directory-enabled networking and operating system support. Hence, 
>> benchmarking
>> and analyzing the performance of a directory server is very crucial. In 
>> this
>> paper, we analyze the performance of the openldap-2.0.23, which is an open
>> source LDAP directory server available from the OpenLDAP Foundation. Our 
>> goal is
>> to benchmark the server performance through a variety of experiments. We 
>> measure
>> the throughput and the latency for LDAP queries using a stress test, where 
>> the
>> server is heavily loaded by a number clients simultaneously accessing the
>> service. In other experiments, we focus on analyzing the update performance 
>> and
>> the effect of indexing on the performance of LDAP queries. Finally, we 
>> study and
>> analyze our experimental results. Our benchmark shows that that the 
>> directory
>> gets saturated around 100 simultaneous accesses and indexing/caching can 
>> affect
>> the latency by an order of magnitude. We posit that our results provide
>> interesting insights into identifying system bottlenecks and tuning the 
>> server
>> performance. Moreover, our results can be used as a guide while developing
>> future directory server implementations.

Dhananjay Kulkarni
email: kulkarni@cs.ucr.edu
web  : http://www.cs.ucr.edu/~kulkarni
phone: 951-787-2522
"If you can't do great things, you can always try to do small things
in a great way"