I have a WCF service that receives requests from a Sivlerlight client, it reads or writes to a SQL database and then it sends the response back to the client.
With 6 or 7 clients delay starts to increase and I would like to check if the more instances I have, the lesser delay (because of load balancing).
I am trying to implement the autoscaling application block but the rules for storage (such as queue lenght) or CPU usage aren't useful for me because I don't use a storage account and although I have 10 clients connected to the service, the instance CPU usage barely reaches 10%.
How could I set a rule to start a new instance when a certain number of connections is reached?