Open Source - Open Cloud?
Cloud services need to be built as “open” from day one
Published 09:53, 17 August 11
I read Simon Phipps’s blog about the OCI with interest. He says, “The Open Cloud Initiative aims to reinterpret the principles of software freedom for a new generation of computing, just as OSI did at the end of the 90’s.”
The key quote to me is: “The mechanisms of source distribution that allow open source licenses to leverage copyright law are largely absent in cloud computing, so OCI aims to “develop and maintain a set of open cloud principles by way of an open community consensus process” as well as to “Persuade organisations and vendors to comply with the open cloud principles.”
Simon accurately observes that it will be interesting to see how the OCI will achieve the last part (the persuasion bit).
Let’s dive into that a bit more. Open source works because it is a win-win scenario. Numerous studies drive this point home and the OSI home page says in the very first paragraph: “The promise of open source is better quality, higher reliability, more flexibility, lower cost, and an end to predatory vendor lock-in.” The vendor gets a better product that is cheaper to develop and support and the user gets a better product that is cheaper to procure and maintain. Win-win.
The Open Cloud Initiative has defined a set of key principles (that come with supporting definitions):
- Interoperability between cloud computing products
- No barriers to entry and no barriers to exit for users of services
- Open process to develop required new standards
These make sense and definitely create the win-scenario for the consumers using cloud services. But where is the win-win? Why would a vendor embrace these? How does embracing the above principles make the vendor service a better one?
No such key principles exist today for the on-premise world. If you want to migrate from SAP to Oracle (or the other way around) you are in for a serious migration effort. Even if the products provide the same type of functionality, they are not compatible on the API or data-level. Most vendors today strive for lock-in through differentiation, the technical/API/data-lock in is most often secondary - even if it creates the bigger barrier many times.
So, where does the win-win come from then?
If the service itself is the commodity, and provided by many vendors, the barrier for new vendors to enter the market is low. Differentiation is then hard and consumer demand will drive out the implementation of the key principles from above. Such a situation is closest at hand in the Infrastructure as a Service (IaaS) layer of the cloud.
Things get more complex when we move up the stack in the cloud to platform (PaaS) and software (SaaS). No commodity is there yet. Most likely, it will never happen in the software space if the same type of very large integrated suites dominate the service cloud as in the on-premise world of today.
But what if the service cloud could really bring home the vision of service orientation? Joe McKendrick, in his July blog says, “We have SOA providing the architecture, governance and orchestration for services delivered through cloud mechanisms, both internally and externally across the Internet.
Readers, is it time to tear down the separate SOA and cloud silos and start addressing service technology as one?” I get what he is saying. What if there are 1000s of smaller grained services out there that a consumer (enterprise) could use? Then the individual weight of a single service provider would not be large enough to enforce lock-in (including those service providers that would handle the flow (process) control across the various services).
If we look at the open source on-premise world, it is evident that all successful projects are, to a high degree, new projects. Very few existing projects have successfully converted to an open source model. One could hence say that the open source on-premise world created a new segment of products (based on the new projects).
If one applies that same analogy to the cloud, it means that cloud services need to be built as “open” from day one. Converting non-open cloud services to open ones later on will be a difficult task.
Tomas Nystrom, Global Open Source Lead, Accenture.