Azure Architectural Pattern – Implementing Master-Slave Model in IaaS

December 9, 2013 Off By David

Grazed from MSDN. Author: Igor Pagliai.

During the last two years, I had the privilege to work with some important Microsoft Global Partners, all over the world, on several important Azure projects. Since I personally like to share my findings and let the broader community to learn from my experiences, in this blog post I decided to discuss an interesting architectural pattern to overcome to one of the most known Azure Load Balancer (LB) limitation, which is the implementation of a MASTER-SLAVE architecture.

In this post, I will concentrate my attention on Azure IaaS Virtual Machines (IaaS VM). I am sure that at many of you already know that the Azure Load Balancer (LB) is great providing load-balancing on incoming connections/requests among a pool of VMs, but what happen if you need:…

  • Only one “active” VM at time (the MASTER) must serves all the incoming connections/requests;
  • The “passive” VM (the SLAVE) must be ready to take over processing only when the MASTER is down;
  • The failover mechanism for routing Incoming connections/requests must be transparent to users/clients and other tiers in the application architecture;

The first idea that probably comes in mind is to leverage Azure LB “Load-balanced Endpoint” and “Custom Probes” but there are at least a couple of problems:…

Read more from the source @ http://blogs.msdn.com/b/igorpag/archive/2013/12/09/azure-architectural-pattern-implementing-master-slave-model-in-iaas.aspx

CategoryNews