Azure Site Recovery Deployment Planner for VMware and Hyper-V

Finally, Microsoft launched the Azure Site Recovery (ASR) planner to help you understand, design and scale your disaster recovery (DR) solutions in Azure for local deployments of VMware and Hyper-V.

Background

The process of enabling replication of a virtual machine in Azure is very simple. Actually, it is the first practical lab that my students do when I come to ASR. What is more complicated is understanding what is already implemented, determining what to implement in Azure, and sizing the bandwidth requirements.

Ah! The question always present in the cloud: "How much bandwidth do I need?" In the case of ASR or any DR solution / service, the only answer is: "You need everything you need." Few are always happy with that answer, as you can imagine. The bandwidth requirements are divided as follows:

•     Initial synchronization: the initial synchronization is when you use the most amount of bandwidth; it will gradually add virtual machines to the ASR service and each VM will completely copy its disks to the cloud.
•     Replication: each virtual machine goes from initial synchronization to replication once the initial synchronization has finished. VMware and physical machines use continuous asynchronous replication and Hyper-V virtual machines use interval-based asynchronous replication based on a policy you define (every 30 seconds, 5 or 15 minutes) and associate it with virtual machines.

The initial synchronization will probably take a long time to complete, from machine 1 to the last machine. How much time is a matter of how much data you need to copy compared to how much bandwidth you have available. In the long term, your bandwidth requirements are dictated by the loss of data; the changes should be replicated soon after (asynchronous replication) and the changes are confirmed in the local storage.

Until now, we did not have excellent tools to estimate bandwidth requirements or other requirements, such as storage accounts (level and performance). A tool based on a spreadsheet called Azure Site Recovery Capacity Planner could take information from other Hyper-V / VMware monitoring tools, but that required a lot of manual effort. I do not know any of my clients who use the tools.

On March 1, Microsoft released a preview for a new tool that was only compatible with VMware, the ASR deployment planner for VMware. This was a tool that scanned / monitored our local environment and produced a report that dimensioned and helped to plan the DR site. Now we had something we needed ... more or less. It was supposed that a Hyper-V launch would come soon after ... and it did not happen that way. Ignite (end of September) came and went too. But in December, Microsoft made the tool generally available with support for VMware and Hyper-V.

What the ASR deployment planner does

The ASR implementation scheduler connects to your hosts to collect data. Note that there is nothing installed in the virtual machines and it is not necessary to implement any ASR component. The data that is collected allows the tool to generate a report with a lot of information and recommendations, which include:

•     Compatibility: Are the configuration of the virtual machine and the guest operating system compatible with Azure and ASR?
•     Bandwidth: the amount of bandwidth required for replication and how this bandwidth affects the desired recovery point objective (RPO: the amount of data lost after failover).
•     Azure infrastructure requirements: the amount and level of storage accounts required to maintain the required performance of the failed virtual machines. It is also provided with guidance on the required number of cores (subscription limits) and series / sizes of virtual machines (optionally configured after the initial synchronization is completed).
•     Local infrastructure requirements: Hyper-V replication is based on Hyper-V Replica, which consumes storage for Hyper-V (HRL) replication log files. VMware customers will receive guidance on the configuration / process server requirements.
•     Initial synchronization: The more machines you add to replication at the same time, the worse the bandwidth will be. The tool will make recommendations about the size of the batch versus the available bandwidth.

   
Cost analysis: an estimate of the costs of Azure will be generated.

The addition of this tool should make it easier to plan ASR implementations, but you should still have a good understanding of:

•       What to replicate versus what not to replicate and what should be implemented with application layer replication over network connections from site to site
•       What you should pre-deploy in the DR virtual network
•       What you should pre-deploy in any sandbox test virtual failover network