The ABN AMRO Morgans site is running hardware load balanced servers (6
in total) that are session aware.  We don't use Cluster Cats.

We use an Array Load Balancer with server cookies enabled.  This means
as the Load Balancer allocates a server it provides its own session
cookie to the browser so that when the user returns on subsequent calls
while the browser is still open, they are returned to the same server.
This ensures that their session information is maintained and that they
don't jump around between servers.

NOTE: we have still found that FRAMES do cause an issue of jumping
around (but believe this to be a browser problem rather than a Load
Balancer or CFMX problem).  This means that you are sometimes prompted
to log in more than once in the Admin tools (as it is partially
frames/iframes based).

We also have modified the login/authentication process for our site
(instance, not core) to ensure session based cookies (i.e. use
<CFCOOKIE> on CFID and CFTOKEN to expire immediately) so that when they
shut their browser down the session is killed.

We don't use FU - but, FarCry now has a config variable for Verity
collections that can be used to provide a central location for them.  We
point all our webservers at a central drive array unit (currently a NAS,
soon to be a SAN) so that we only update on one place.

Ummmmm - what else.....

Oh yes, with your APPS.CFM - as well as an entry for the application
domain itself, we also placed additional entries for each servers IP
address so that we can maintain the site directly from any of the
servers.  This helps when you are trying to debug at the server level.

Also, we are doing all our background task processing on one machine.
This is actually an issue for FarCry/CFMX.  The tools in FarCry are not
multi-machine aware - so if you go to Admin via the Load Balancer, you
wont know which machine the task has been created on.  Even worse, next
time you go back you may not be on the same machine and you tasks may
appear to be gone.

So, there are management issues with it - but none that you would not
have in a multi-machine CFMX (even PHP/ASP) environment anyways.

Regards
Gary Menzel


> -----Original Message-----
> From: [EMAIL PROTECTED]
[mailto:bounce-farcry-
> [EMAIL PROTECTED] On Behalf Of Michael @ Four Eyes
> Sent: Saturday, 21 February 2004 1:33 AM
> To: FarCry Developers
> Subject: [farcry-dev] Re: Clustering?
> 
> The gold bank project is clustered.   What are you interested in
knowing?
> 
> Basically, they have 2 web servers, talking to the SQL server.  They
are
> using a sched task to keep the FU and verity synced on both boxes and
a
> .bat
> file to look at the images and files directories to keep them synced.
The
> images and files syncing is clumsy right now, so we are planning to
create
> storage server that both boxes can upload and read from.
> 
> 
> --
> Regards,
> Michael J. Sammut
> ________________________________________________
> F O U R  E Y E S  P R O D U C T I O N S
> 
> think | plan | create :: web site design & development :: NYC
> 
> E. [EMAIL PROTECTED] | T: 718.254.9557 ext. 101 | F: 718.254.0399
> 
> W. http://www.foureyes.com
> "Nathan Mische" <[EMAIL PROTECTED]> wrote in message news:[EMAIL PROTECTED]
> dev...
> 
> 
> Has anyone scaled farcry to multiple servers? Is it even possible?
> 
> --Nathan
> 
> 
> 
> 
> ---
> You are currently subscribed to farcry-dev as:
[EMAIL PROTECTED]
> To unsubscribe send a blank email to leave-farcry-dev-
> [EMAIL PROTECTED]
> 
> MXDU2004 + Macromedia DevCon AsiaPac + Sydney, Australia
> http://www.mxdu.com/ + 24-25 February, 2004


---
You are currently subscribed to farcry-dev 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