[oclug]PostgreSQL Hardware

[Milan Budimirovic]

Brad Barnett wrote:

> 
> Heh!  Yes Milan, it was the introductory sentence to the point you were
> making.  That's the whole point.  By putting 'not necessarily' as the
> first sentence, you indicate that 'not necessarily the above', followed by
> your reasons why!

Which is the opposite of what you accused me of in your previous post.
Please make up your mind. No, wait. Scratch that. I don't want to
encourage you.

> 
> That being the case, you were contradicting what I said.  Such as, 'You
> used "not necessarily" against statements which I summarized above.'

Once more from the top. An argument is a set of related statements
intended to establish a proposition. It's not just contradiction...

This is inane. You have me quoting Monty Python now.


> 
> Let's stop with the games man! ;)

I will if you will.


> > In any event they typically involve very
> > simple queries, and the complexity is in how the data is structured and
> > indexed.
> 
> Typically?

Yes. The vast majority of Google users do simple searches on the main
page.

> I'm not talking about the law of averages.  You seem to be
> though.  We're talking about one server.  My statement is that until you
> know what he is doing with that server, you can't make decisions on what
> hardware will be needed.
> 
> You seem to be arguing that you can, based on what the "average" or
> "typical" use for a server would be.  Do you honestly want to continue to
> claim that you would buy and configure a database server, without even
> knowing what it was for?  When it could be for an online store with
> thousands of customers, or a search engine designed to search research
> papers?
> 
> Dude, if you want to tell me that you'd buy and build a database server
> blind, then fine by me.  Don't expect me to though!
> 

No, what I'm saying is that it's unlikely that he will need more
hardware than what he has proposed. That if he has a performance
bottleneck it is more likely to be the web server or the bandwidth, and
that if the database really is slow he can try things like indexing
columns and adjusting environmental variables, before throwing more
hardware at the solution.

> 
> I'm glad to see you using 'almost' again.  We aren't talking, as mentioned
> above, about the average.  We aren't talking about the common denominator
> here.  We're talking about a specific system, and again, you seem to be
> arguing that you can somehow magically build database servers, without
> knowing what the client is going to use them for.  Just because it is a
> database server to be accessed by a web page obviously does not indicate,
> with certainty, what it's load will be.
> 
> Your own "almost always" above states that you agree with me, otherwise it
> would be an "always".
> 

By your logic, if I can't say anything with absolute certainty I can't
say anything at all. Which is nonsense.