[Openroad-users] (no subject) - OR App Server Architecture Design

[Jonathan Barton]

Thanks Pete.
 
Our setup sounds very similar to yours and to implement the the failover
as you suggested would not mean a great change for us: a common library
class and a config change.  Seems to be the best of both worlds.
 
Many thanks.
 
Jonathan

________________________________

From: openroad-users-bounces at peerlessit.com
[mailto:openroad-users-bounces at peerlessit.com] On Behalf Of Pete
Rabjohns
Sent: 19 September 2007 10:05
To: International OpenROAD Users
Subject: Re: [Openroad-users] (no subject)



Hi Jonathon,

 

We wrap the app server calls into a separate layer on the client and
build the resilience and failover into the client code, so that if it
cannot connect to the primary app server it fails over to a secondary
app server. It's not a perfect solution because it performs the failover
on a per-call basis. The client code has a list of app servers it can
connect to in a config file. Because each call is stateless then it
reads the config file per call.

 

If you wanted to make this a little more intelligent, the client app
could reconfigure on the fly, so that when one goes down it 'remembers'
for a specified period of time, until switching back.

 

HTH

 

Pete

 

________________________________

From: openroad-users-bounces at peerlessit.com
[mailto:openroad-users-bounces at peerlessit.com] On Behalf Of Jonathan
Barton
Sent: 19 September 2007 09:50
To: International OpenROAD Users
Subject: [Openroad-users] (no subject)

 

Hi all

 

I am looking for a bit of advice OR App Server architecture deisgn.

 

We have two web servers (resiliance/load balancing with .net Web
Services) both of which communicate with an OpenROAD Application Server,
and all is well.

 

I have been testing the use of an additional OR App Server using remote
nameserver entries on the primary OR App Server for resiliance purposes
and I am ready to deploy the new server to the live environment.  But I
am concerned about practicalities of effective resiliance and software
releases requiring down-time.

 

Is there any benefit of not using the remote nameserver entries on the
Primary OR App Server, but rather have a designated App Server for each
web server?  This would enable us to deploy new software without taking
the system off-line by simply draining a web server from our nlb
(Windows 2003 Network Load Balancing).  Although this would not be true
load balancing as far as the App Server is concerned.

 

Any experience shared would be greatly appreciated.

 

Many thanks.

 

JONATHAN

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

"University of the West of England"

* Work Email:         jonathan.barton at uwe.ac.uk

* Work No:             0117 3281075

 

P Please consider the environment before printing.

 

________________________________

This email was independently scanned for viruses by McAfee anti-virus
software and none were found 

________________________________

This incoming email to UWE has been independently scanned for viruses by
McAfee anti-virus software and none were detected 



This email was independently scanned for viruses by McAfee anti-virus software and none were found
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.peerlessit.com/pipermail/openroad-users/attachments/20070919/a451ad41/attachment.html