In his session at TechEd NA 2013 yesterday, Ben Armstrong brought up an important point with cloud computing. People today, ask themselves if they should deploy a private cloud or subscribe to a particular public cloud. This is a “one or the other” decision in their minds. And the public cloud in their minds is usually just one public cloud. Ask Netflix what they think of limiting themselves to just one public cloud (Amazon in VA, USA with the uptime of a …. you catch my drift).
On the Microsoft stack, it can be a different question: do I deploy this VM/service on a public cloud or do I place it on my private cloud. And if it’s a public cloud, which one? Do I use the elasticity and scale of Azure, or do I use the local hosting company that can offer engineering time?
Here is the usual marketing slide at the start of any recent Microsoft presentation on cloud or virtualisation. Does it mean anything to you? Fluff, right?
Maybe this cloud stuff isn’t so fluffy! I presented the following at E2EVC in Copenhagen last week. This diagram is based on what you can do right now with Azure, Windows Server 2012, and System Center 2012 (with an NVGRE gateway in the hosting company).
The hosting company has a multi-tenant cloud, using Hyper-V Network Virtualization (HNV) to simplify the physical networks, and provide tenant isolation. VMM manages the cloud(s). SPF resides in-front of a HA VMM installation, and provides a hosting API.
Coca Cola *ahem* is running a private cloud. That’s also managed by VMM. Users of this cloud use the portal provided by App Controller and can deploy services on this private cloud, availing of the security, SLAs, custom engineering, etc that they expect from IT.
The cloud admins have created a contract with the hosting company, enabling App Controller to interact with SPF in the hosting company. Now the users in Coca Cola can deploy services on-premise or in the hosting company. It’s no longer an exclusive-or (XOR) decision when placing services, it has become an and/or decision. Look at the diagram; Pepsi and Lidl *ahem* are also in this hosting company’s public cloud, completely isolated from Coca Cola thanks to HNV.
Coca Cola’s cloud admins have also created a contract with Azure, and App Controller can reach into there too. Now the self-service users have a choice when deploying a service:
- Use the highly secured and managed on premise cloud
- Deploy services into the more scalable but local hosting company with a customised services contract
- Or use the instantly scalable and elastic Azure IaaS service
And it’s not just a placement decision. You can re-evaluate placement and move virtual machines (offline) from one cloud to another. There is no migration, no OVF to mess with, etc.
This is the benefit of having 1 consistent platform, with Windows Server and System Center running in all three environments.
For the hosting company, they aren’t left in the cold against Azure. Windows Azure Services for Windows Server (aka “Katal” and vNext is called Windows Azure Pack) provides the exact same portal for their own use. It can be skinned and customized. Now the hosting company can provide the very same services as Azure, using the very same Hyper-V and HNV as is used in the Microsoft public cloud.
Now look at that slide again. It’s not marketing fluff anymore. It’s one stack, designed from the ground for this purpose, in many clouds that empowers you with the ability to choose.