I'm a little bit confused by how scaling in Azure works. I'm using a Cloud Service and have 2 web roles running a PHP application. I can RDP on both machines and both applications run great on each machine. Also I don't have any problems calling the staging URL.

But I can't figure out if I configure scaling so that 2 machines run always, if I have to configure a load balancer somehow. Or is this already done for me?

In Azure VM's I had to create a load-balanced set endpoint for an endpoint, but what about cloud services?

And how is this done in the XML configuration file for my service? What if I don't do it?

Hi,

Scaling is affected by core usage. Larger role instances or Virtual Machines use more cores. You can only scale an application within the limit of cores for your subscription. For example, if your subscription has a limit of twenty cores and you run an application with two medium sized Virtual Machines (a total of four cores), you can only scale up other cloud service deployments in your subscription by sixteen cores. All Virtual Machines in an availability set that are used in scaling an application must be the same size.

Windows Azure supports load balance for cloud services and standard websites, we just need to set instance count to more than 1 to enable load balance. For virtual machines, it needs to set up manually.

Please refer this link for Load Balance a Virtual Machine:

http://www.windowsazure.com/en-us/manage/windows/common-tasks/how-to-load-balance-virtual-machines/  for more information.

Auto scale lets you set scaling limits and scheduling goals to ensure you are always getting optimal performance

Please refer this link for Scaling on Cloud Services: http://azure.microsoft.com/en-us/services/cloud-services/

Also, Please refer this link for Scaling an Application : http://azure.microsoft.com/en-us/documentation/articles/cloud-services-how-to-scale/

XML configuration : Azure (Load-balanced) Endpoints can only be used for TCP/UDP based services. please check https://techlib.barracuda.com/display/BNGv54/How+to+Configure+a+High+Availability+Cluster+in+Azure/printable for the detailed information

Hope this helps.

Regards,

Shirisha Paderu.

• Edited by Shirisha PaderuMicrosoft contingent staff, Moderator Sunday, July 27, 2014 2:47 PM
• Proposed as answer by Shirisha PaderuMicrosoft contingent staff, Moderator Sunday, July 27, 2014 2:47 PM
• Marked as answer by JWAzure Monday, July 28, 2014 7:49 AM

Thank you for the detailed answer.

"Windows Azure supports load balance for cloud services and standard websites, we just need to set instance count to more than 1 to enable load balance. For virtual machines, it needs to set up manually."

That's what I wanted to hear.

I am still however not quite clear. You are describing load balancing on input end points only, right? 

If my cloud service has 2 web roles, one with input endpoints and hence publicly available, and the other with internal endpoints, then role on internal endpoints will not be load balanced at all.

I know I can add internal load balancer through power shell and add my VMs running internal end points to it. But if I now auto scale, my internal load balancer will make no use of those new VMs generated for my role.

Is there a way to make internal load balancer "aware" of auto scaling and have it add/remove VMs as they are added/removed for my second web role?