We are assisting a customer with their first Azure setup which is an Azure A3 server in Northern Europe running Windows 2012 R2, II7 and SQL along with their own application.

The customer actually has three servers, live and testing on Azure and a development server hosted in the cloud at Bytemark. The server at Bytemark is we've been told lower physical spec for memory and CPU running the same services and applications on Windows 2008.

The customer has not gone live with their services on this solution yet but have raised some concerns about bandwidth observations on the Azure servers in comparison with those on the Bytemark servers.

I believe the customer has used a basic "Broadband Speed Test" (we believe testmy.net) from the console of the virtual server and have come away with the following figures.

Bytemark Server: Downstream Average: 32 Mbps Bytemark Server: Upstream Average: 44 Mbps

Azure Server: Downstream Average: 103 Mbps Azure Server: Upstream Average: 6 Mbps

Could I please ask if these are they type of figures you would expect to see from such an Azure Server. Does the bandwidth performance of servers increase with their specification D having greater bandwidth than A and so on?



While there is no exact answer for you, consider this:

  • A-series is the oldest, and I would strongly recommend you benchmark with D-series.
  • Bandwidth to a VM is independent of bandwidth to Azure Storage. So, if you are using your VM's to deliver static content (videos/images/etc), you should also benchmark these assets directly with blob storage, which has independent max of 60MB/s throughput per blob.
  • There are no exact benchmarks published for the VM NICs. So you will need to benchmark the various VM sizes to draw your own conclusion.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.