2

I am now learning to deploy the web server on the cloud such as amazon or rackspace.

yesterday, I found a configuration of servers on this rackspace link: http://www.rackspace.co.uk/calculator. in this GUI, apart from adding the normal servers, they can additionally provide a load balancer server, but this load balancer server is much more expensive.

Now I want to use the AWS to build such configuration, for example, 1 or 2 normal application servers + 1 load balancer. the load balancer will be configured with the nginx. what kind of requirement of server for that load balancer in terms of size of disk, CPU, size of memory, if I want the load balancer to handle 500 / 1000 concurrent request? how do you calculate it ?

and moreover how to consider the websocket connection through nginx? at last, for prototyping the multiple apps, putting the load balancer server with application server in the same physical hosting server, I think it is possible, but would both load balancer and application server will compete for the resources on the server , is it ?

user824624
  • 7,077
  • 27
  • 106
  • 183

1 Answers1

0

Calculating the capacity of a virtual load balancer is on a case by case basis. It depends on your access patterns and data size. So the only real way is to run benchmarks on it till it starts to slow down, then determine which bottleneck it's hitting (CPU/Memory/Iops/Network) and adjust your instance type accordingly. Tools like siege (http://www.joedog.org/siege-home/) will help you find out what your limits are.

However, since you are just starting out, I would suggest using AWS Elastic Load Balancers instead of creating your own Nginx EC2 based load balancer. They are a lot cheaper and take the hassle and let you get back to working on your application.

Anthony Hobbs
  • 48
  • 2