[thesite] blockfactor revisted

Daniel J. Cody djc at starkmedia.com
Fri Oct 26 02:29:37 CDT 2001


.jeff wrote:


>>>where did i say i was blaming them on us.  i'm just
>>>trying to explore the issues to try to find some
>>>explanations to the things we're seeing.  that's all.
>>>
>>exactly. we can code our CMS perfectly. the fact is,
>>the performance issues aren't our fault. they're MM's
>>fault(4 pages/sec? oy.)
>><><><><><><><><><><><><><><><><><><><><><><><><><><><><><
>>
> 
> actually, our cms has a lot of inefficiencies currently -- things that i've
> only found out *since* building v2.0.  it's far from being perfect.  4
> pages/sec are our fault until we've combed the code for the problem areas.


ok, so lets fix it. according to the MM whitepaper, we'll be able to get 
up to 6p/sec with a 4 way box(generalization). weee! a whole extra 2 
pages/sec


> speaking of that, as you see reads of session variables (session.user.priv,
> for example), change them to have app. prepended to them
> (app.session.user.priv, for example).  that way we don't have to lock to
> read them since i'm doing a lock once and duplicating the session scope out
> to a structure without locking issues.
> 
> writes to session variables are a different issue.  we'll need to lock those
> writes and also write to the matching app. copy.


IMO, no matter how tightly we code our CMS, it won't scale. unless we 
have a couple 4 way boxes.

 
> i can go into a long diatribe about my thoughts on the open source vs.
> commercial software situation.  i don't think this is really the place for
> it.


i'd like to hear it. post it here :)

 
> if you've ponied up the money for oracle how much difference would another
> couple grand honestly make?  i'm not saying it's right, just how i see it.


we didn't pony up for oracle. and a couple grand does make a difference 
when you're a NFP.

.djc.






More information about the thesite mailing list