Understanding when a hybrid cloud really makes sense: 3 real world examples

“How about a couple of real world examples of situations where a Hybrid cloud makes sense?”. Seems like a simple enough question to be asked by a non technical colleague. I have worked with a lot of different groups and customers who have put together a number of such scenarios to give me enough background to come back with a couple of possible options.

Before we go there – some quick review: If you are looking for a good range of background info on Azure and Hybrid Cloud – I strongly suggest you check out some of the MicrosoftVirtualAcademy Hybrid Cloud Training resources.

How about we keep this simple and focused on a couple of quick hit scenarios where this played out.  Remember, as you have read over the last couple of months – you have multiple methods of connectivity back into your Azure v-nets to get a hybrid solution – here is a quick recap:

• Point to Site: quick and dirty, easy to setup.  Down side – only one box “on premises” will have access to the Azure environment.
• Site to Site: Software or Hardware gateways supported.  Much better and “production ready” based on your comfort level.  Software based gateways running Windows Server 2012 / 2012 R2 can meet your needs OR a supported hardware device like a Pix firewall if you prefer a hardware solution. Advantage over Point to Site – you define the subnets that have connectivity to the Azure V-Net and as many machines as you like can now securely access the Azure environment.
• ExpressRoute: a service provided by one of our partners who probably already provide you WAN connectivity to your branch offices.  Very fast, reliable and low latency. You are literally patching in an Azure vNet like it’s a branch office.

I went from cheapest to most expensive as well as least scalable to most scalable.  With that in mind – lets get back to those three quick hit scenarios:

1) Keep my data on-site, in my datacenter.

I get this one all the time. People are ready to work with a public cloud offering, but want to ease their way in with regards to maintaining control of their data. If your applications can handle being separated from the data back ends and you have available bandwidth to support whatever latency they can work with – this is a simple case for a hybrid solution.

2) Fire up a quick pilot without capital costs.

Say your virtualization hypervisors are all in production and you want to spin up a couple of app servers or setup a trial workload from some vendor.  When you have spent the time to understand how easy it is to establish a Hybrid connection to Azure, you now have “Subnet(s) on demand” with whatever compute power you need – all being billed only for the time they are ON, broken down to by-the-minute charges.  No licensing to worry about, no delay to get hardware spec’ed, ordered and installed, no additional physical demands on your infrastructure.  You also get to determine which portions of your network have access to this secure subnet (no public access to it from the internet unless you open it up). When the pilot isn’t running (off work hours) you shut down machines to suspend billing and start them up when you need them again.  Once the pilot has been completed – you move machines over to on-premises hardware (they are VHD files) or establish a “production” subnet on demand.

3) Disaster Recovery for certain workloads.

I say certain workloads because some workloads are supported in Azure, while others aren’t supported at this time.  But – for those that are, you can establish a Site to Site VPN connection and use a service like Azure Site Recovery to replicate select on-premises servers up to Azure and keep them synchronized.  If you need to do a DR test (or the real thing) you initiate a Site Recovery and Azure handles ensuring the last replication finishes (where possible) and brings the systems up in the properly defined order.  If you have your hybrid connectivity established to your other sites – they would have connectivity to the DR site and services that had been temporarily re-located there by ASR.  This is by far the most complex of the three scenarios – but it’s just damn awesome in its powerful capabilities.

Those are my quick three scenarios where Hybrid cloud solutions just make sense.  What do you think – have you come across any that didn’t make my top 3 list?