When we say cloud, our mind is flying to scalability. Yes, this is the beautiful of cloud. In the last period of time I had the opportunity to work on a project that needs to scale from 2-4 instances to 200-300 instances in a few hours. The peek is pretty short (2-4 hours)
In this kind of moment you realize how hard would your life be be if you would not have load balancer and traffic manager. On on-premises solutions I see almost every month a possible issues reported by customers or external team that load balancer is not behave as expected. Testing a load balancer is pretty hard when you need to simulate 20k-50k clients with different IPs and configurations – and also very expensive.
Next, you realize that using only one data center is not a good idea, you want to be protected if something happens with that data center. Not only this, but because of the client business, we could easily group the load in 4 geographical regions (America, Europe, Asia and Australia). Because in this moment we don’t have a data center in Australia, we decided to go only in 3 data centers and in future will see :-).
With Traffic Manager it will be extremely easy for us to redirect client request to exactly the data center that has their data, without having to redirect them to another nodes(in 80% of cases this should be applicable for us).
What next? Additional to this, we need a mechanism to cache clients request until our instances are ready. For this purpose, Azure Service Bus is ready to help us. Once the command is send to our system, clients subscribe to Service Bus and wait a notification from backend. Also we are using Azure Service Bus to distribute the work to multiple instances.
And yes, we were able to migrate a system from on-premises to cloud with only a few modification. The legacy part of the system was not modified and works great. This is the beautiful of Microsoft Azure.