Thanks Scott.  That clears that up then!

It must be a difficult thing to figure out - how much grunt to give a site
like that -  where you have no benchmark to start from.  Somewhere along the
line you have to decide how many simultaneous sessions to cater for, and in
the case of Jetstar, it must have been a total guess.  They'd have known the
publicity and the $29 fares would have generated a lot of traffic, but how
much?  Putting a number on it must have been tricky, and judging by the
results, not an exact science.

Then again, it could be a fair decision to make that there will be a big
peak at the launch, which will settle down to a lower level later.  So to
engineer the servers for the large launch traffic at the start, and then be
saddled with the costs of all that grunt that will never be used again would
generate debate in the launch planning I'd have thought.


Cheers
Mike Kear
Windsor, NSW, Australia
AFP Webworks
http://afpwebworks.com


-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Scott Barnes
Sent: Sunday, 29 February 2004 4:16 PM
To: CFAussie Mailing List
Subject: [cfaussie] Re: So who designed the Jetstar site?



Michael Kear wrote:
> 
> I would love to know more about the server architecture decisions on this
> site - how they worked out what the load might be and how they scoped it
for
> that load.     Does anyone have any input for that?


The decisions were made at a sleazy hooters bar one night after a few 
too many drinks. The way they worked out the load was, to take a packet 
of tooth picks, throw them in the air, and the ones that land on the 
table, was the expected total number of hits likely to visit the site. 
They then outlined a prison break style mud-map on a cocktail napkin, 
followed by some outlandish purchasing from Dell online.
[snip]



---
You are currently subscribed to cfaussie as: [EMAIL PROTECTED]
To unsubscribe send a blank email to [EMAIL PROTECTED]

MXDU2004 + Macromedia DevCon AsiaPac + Sydney, Australia
http://www.mxdu.com/ + 24-25 February, 2004

Reply via email to