Todays blog post was inspired by Red Hat’s Vice President for cloud business, Scott Crenshaw, and his definition of an open cloud:
- Open source
- Viable, independent community
- Based on open standards
- Unencumbered by patents and other IP restrictions
- Lets you deploy to your choice of infrastructure
- Pluggable, extensible, and open API
- Enables portability across clouds
Although I think this is a very good start for a discussion, I do not fully agree with his definition!
Open standards, APIs, and portability
I don’t doubt these points of Mr. Crenshaw’s definition — I see them as the most important criteria for a cloud to be called open. Cloud consumers should be able to seamlessly move their workloads from one open cloud to another. There is no room for vendor locking — I fully agree here with Mr. Crenshaw!
Open source, independent community and patents
Considering the fact that Mr. Crenshaw is a Red Hat employee, no wonder, he sees open source as a requirement for a cloud to be open. However, is that really the case? I doubt it.
Sure, open source software and the viable, independent communities have their benefits, but that is not specific to cloud computing, nor is it a requirement for an open cloud. I honor the fact that open source-based software stacks — such as OpenStack for example — are implementing open standards and interfaces, and are driving their definition. But after they are established, I see no reason why a closed source software — that complies with those standards — should not be considered open.
Traditionally software products are within the responsibility of the IT departments. Cloud computing changes this paradigm to a certain extent. With cloud computing, we often see a direct relationship between a business unit and the cloud vendor bypassing the IT department. Now, we can argue whether this is good or bad, but what we do need to pay attention to is seeing the product from a different viewpoint. The pure technical aspects become less important. So, if it is less important whether or not the cloud is based on open source software, the important question is: what can it do and not do?
Choice of infrastructure
I admit that the open choice of infrastructure can help eliminate vendor locking. But I personally consider the support of different platforms and infrastructures as a feature and nothing more. Of course, when selecting a cloud software stack or vendor, the provided features must fit to the requirements. And, the more flexible the features are, the more future-proof my selection might be, but that’s not a criteria for an open cloud — at least not for me.
Summary
An open cloud must stick to open standards and implement open interfaces and APIs. I see those as the main criteria for an open cloud. Open source is definitely helping to push these criteria, but is not a mandatory requirement. At the end of the day, cloud consumers must be able to move their cloud workloads and data from one cloud to another, that’s what makes the open cloud reality!
Sources:
“What’s an “Open Cloud,” Anyway? Red Hat Says It’s Not VMware“ by Joe Brockmeier (http://www.readwriteweb.com/cloud/2012/02/whats-an-open-cloud-anyway-red.php?sf3468400=1)