Gartner predicted 75 percent of large enterprises would have hybrid cloud deployments by the end of 2015.
And while we still don't know if this prediction panned out, spending on hybrid clouds is growing at a compound annual rate of 27 percent, on pace to reach $85 billion by 2019 from $25 billion in 2014, according to researcher Markets and Markets.
What does “hybrid cloud” mean? The term has been around for some time. The National Institute of Standards and Technology defines hybrid cloud (pdf) as “a composition of two or more clouds (private, community or public) that remain distinct entities but are bound together, offering the benefits of multiple deployment models.”
This means having some infrastructure in the cloud and other infrastructure in an on-premises corporate datacenter, with potential co-mingling of these two separate islands of infrastructure.
A Blended Infrastructure
Imagine if a hybrid car was built that way: The electric infrastructure of the car would be a distinct entity from the internal combustion infrastructure instead of a single infrastructure that makes both of these systems seem like one system. Hybrid cars have been successful because they blended the electric infrastructure so that it appears to be part of the internal combustion infrastructure — the driver doesn't know the difference.
Based on my experience working with companies as they adopt hybrid cloud, my opinion is that for the cloud to become a useful part of the enterprise, it has to look, act and feel like it is just another room of the corporate datacenter — not a separate infrastructure in a separate facility hundreds of thousands of miles away.
Sound complicated? It’s much easier than you might imagine if you don’t think of the cloud as an entirely separate infrastructure.
5 Tips for a Seamless Cloud Experience
Here are the five key requirements to make the cloud a seamless extension of the corporate datacenter:
- Create a single network: Everything in the cloud needs to look like it is a part of the datacenter. From cloud compute instances to software licensing servers, everything needs to be on the same network as the datacenter. In other words, the datacenter and the cloud should seem like one contiguous entity.
- Automate and orchestrate: While automation is useful in a datacenter environment, it is imperative in a cloud. Most cloud were really built for developers, not IT. The use of cloud automation tools, such as Cloud Formation Templates in AWS, enable IT to simplify interaction with the cloud.
- Co-locate data and apps: Make sure data is close to those who are actively using it. This means co-locating data next to users/applications in the datacenter and also next users/applications in the cloud. Having a global file system with instances in both the datacenter(s) and cloud ensures great performance in both locations without trying to copy the data back and forth, which can cause version control issues.
- Stay safe: Security is probably the most debated aspect of the cloud. While you want to have data next to cloud compute instances for performance reasons, you need to make sure the data is secured properly. Repeat the following term three times: FIPS 140-2. Cloud providers have FIPS 140-2, but you cannot just rely on cloud providers for security as they would need to comply with a government subpoena for your data and not let you know. Your data needs to be secured in your datacenter, in transit and in the cloud with FIPS 140-2 where you own the keys.
- Authenticate: There is no reason to try and federate or replicate Active Directory into the cloud when you can just use your current Active Directory implementation that sits behind your firewall. Enough said.
What Goes Where?
So now that you’re ready to set up a hybrid cloud infrastructure, what enterprise apps do you run in the cloud and what in a datacenter?
In most cases enterprise applications developed for the corporate datacenter should be able to run without change in the cloud and conversely, enterprise applications developed in the cloud should work as is in the corporate datacenter.
However, hybrid cloud exists for a reason. Some things are better suited to move to the cloud and some you might want to keep on premises. Enterprise applications developed for the datacenter that are great candidates for the cloud include: anti-virus scanning, rendering, simulation and search/indexing. Applications that are better candidates to live in a datacenter include financial or medical applications that you might feel more comfortable keeping behind your company firewall using private cloud storage.
While the debates will continue about which applications to run in the cloud and which applications should stay in the corporate datacenter, enterprises need the flexibility of running an application in the cloud, in the datacenter, or in both places. During Oracle's recent launch of new cloud services, Larry Ellison discussed moving an application from on-premises into the cloud and from the cloud back to on-premises. He said, “You can move in both directions.”
Despite the fuzzy definitions and the continued hype, the hybrid cloud will prove its value when enterprise customers just see the cloud and the datacenter as one seamless environment.
Following the five requirements above will help you reach that place where it will be hard to determine where the datacenter stops and the cloud begins.
Title image Lionel Abrial