Re: 8 hours tests ends with inconsistent DB.

[Tony Earnshaw <tonye@billy.demon.nl>]

Wesley D Craig wrote:

> I certainly see your point, above, but I don't think you're going to get 
> much of an idea of scaling using the approach you're using.  As I 
> understand your method, you're trying to put a server up and hammer it 
> without knowing much about configuring it.  I don't think this tells you 
> anything about scale.  This *might* tell you something about how easy it 
> is to give the application to someone who doesn't know anything.  The 
> test you're performing might be appropriate if you're hoping to 
> repackage and distribute OpenLDAP to 50 million customers.
>
> As I understand your *actual* need, you're planning to deploy an 
> infrastructure of some size to serve 50 million customers.  This is a 
> much different task.  In order to do this successfully, you will need to 
> know how many machines to deploy.  You'll need to know how to operate 
> OpenLDAP, including how to configure it.  In order to know that, you'll 
> need to know what the real read & write load is likely to be, average & 
> peak.

Shhh ... ;) This thread has been enormously interesting. It *has* 
exposed at least one Openldap weakness (index corruption), *has* 
demonstrated the need for correct database configuration and log file 
administration.It'd also point toward the use of correct hardware for 
the job in hand. Let it run its course, please?

--Tonni

--

We make out of the quarrel with others rhetoric
but out of the quarrel with ourselves, poetry.

mail: tonye@billy.demon.nl
http://www.billy.demon.nl