The Truth About the Open Cloud: What Really Matters
There is always a lot of talk about the word "open" in tech circles, especially when it comes to the cloud. After all, no one wants to be locked in and shackled to a "closed" system, right?
On Monday, Sept. 9, my friend James Maguire, managing editor at Datamation, is hosting a Google Hangout on the topic of Open Source Cloud, and it looks to be an interesting event. He's got people from Eucalyptus, Red Hat, Nebula and Citrix all set to speak during the hangout.
I've sat on a number of panels in recent years talking about the open cloud, and one question always comes up: What exactly is the open-source cloud anyway?
Simply using an open-source technology does not make a cloud open in my opinion, and it's an argument that I've made every time anyone ever asks me. I like to think like a user, and frankly users in the cloud (and elsewhere) typically have one primary concern—and that's about data.
Is the data portable? That is, if I have my data in one cloud, can I get it out and move it to another? It's a simple question, but one that doesn't have a simple answer in every case.
Going a step further, is the workload portable? That is, if I'm running an application workload on a certain cloud, can I move it to another of my choosing (and not just one operated by the same vendor)?
Yes, open source is important. It means that as a user (if I have the skills) I can bend the code to my will to "scratch my own itch" and get what I need. It also means that real standards can emerge that are supported across multiple vendors. The reality, though, is that many users will want fully baked solutions from vendors. Take OpenStack, for example. If you run a workload in one OpenStack cloud from a vendor (Red Hat, Rackspace, HP, SUSE, Canonical, Piston, Nebula or otherwise), can you easily move your data across those implementations in a fully interoperable way?
The answer today is yes and no. With the OpenStack Image Service (code-named Glance), I should be able to snapshot an image on one implementation and move it to another. That's a form of open that isn't just about open source code. There are other scenarios where users might need or want to have a hybrid deployment moving data across all manner of implementations and vendor solutions, both proprietary and open-source-based.
The question that needs to be asked by cloud consumers isn't just whether or not code is open source (which I personally think it should be); the bigger question that needs to be asked is, Is the solution interoperable? If for some reason a certain solution is not, then even if the code itself is open, the risk of being locked into a solution might very well still exist.
Sean Michael Kerner is a senior editor at eWEEK and InternetNews.com. Follow him on Twitter @TechJournalist.