Before implementing SAP HANA system replication between virtual machines, you must set up and configure your Azure Infrastructure. Broadly speaking, this involves deploying an internal load balancer between multiple VMs in an availability set. One option is to use a preconfigured Azure Resource Manager template available from the Azure marketplace. Here I have an SAP 3 tier marketplace image where all I have to do is fill in the blank parameters. So that's a resource group, which I'll create now. The OS will need to be changed to Linux, in this case, Suse Enterprise Server. Obviously, the database will be HANA, and system availability is high - ok, that's it.

No, seriously, let's go through setting up an internal load balancer through the portal. Firstly, let's create a virtual network called hana_vnet within the hana_rg resource group. The vnet will have two subnets, each of 255 addresses, but you can reduce that number for the back end if you like. I won't enable any security features at this time. That's validated successfully, so I'll hit create. Next, we create an availability set within our resource group and in the same region as the network. Now the virtual machines. I'll create two, hana1 and hana2. They will be identical, using the Suse Enterprise Linux for SAP image and belonging to the availability set we just created. As this is a demo of configuring the network and load balancer, I won't bother adding data disks. Of course, carrying on and installing Pacemaker clustering will require the VM's to have disks attached. Under networking, choose the virtual network we've just created and the annoyingly misspelled backend subnet. The machines behind the internal load balancer aren't exposed to the internet, so they won't have a public IP address. Everything else can remain as default.

Once the VM's have been provisioned, we can create the load balancer. It will be deployed into the same resource group and region and is internal. Use the standard SKU. It has more features, better performance, and requires less configuration. Add a front end IP, selecting the virtual network's front end subnet, making the assignment static, and give it an internal IP address. Next, we add the VM's to the load balancer backend pool by clicking the add button and then choosing the virtual machines.

After the load balancer has been provisioned, we can set up the health probe by going into heath probes under settings. Give the health probe a name, leave the protocol as TCP and change the port to 62503. Interval and unhealthy threshold can remain as is. Click add. The last thing we need to do from an infrastructure point of view is add a load balancing rule. Here we select the various resources we've just set up. Choose the front end IP address from the drop-down and check HA ports. Select the backend pool and the health probe. Finally, make sure floating IP is enabled and click add.