Deployment Architecture for Nginx Reverse/Caching/Load Balancing with Puppet

[jjarokerno]

Hello,

We are installing nginx for the first time.  Could someone please
comment on the deployment architecture we selected.

Is the architecture illustrated below the best one to meet the following
objectives:

	- A physical host with a single public IP address must be virtualized
into many virtual servers.  These virtual servers run Apache and must be
reachable from the single public IP.
	- Static content from the virtual servers must be cached
	- One of the virtual servers runs Puppet server.  We wish to optimize
Puppet server performance by using caching and load balancing of the
backend Mongrel servers.
	
The architecture to meet these objectives is illustrated below.  In this
design:

	- Nginx for reverse proxy and caching proxy runs on a virtual server
and answers all requests from the public IP address.  As a reverse
proxy, nginx routes requests to the appropriate virtual server.  It also
serves static content from these virtual servers.
	- On the virtual server running puppet, we use a second instance of
nginx, which is configured with the Fair Load Balancer to serve requests
to the Mongrel servers.
	
Is this the optimal configuration of nginx for these objectives?  Could
the above objectives be met with only a single instance of nginx?

[img]http://jaroker.com/nginx/nginx_deploy.png[/img]

Any comments or suggestions would be greatly appreciated.

Thank you,
Jon

Posted at Nginx Forum: http://forum.nginx.org/read.php?2,154806,154806#msg-154806