It is indeed some time since OPENi has finished, and I think it is time to evaluate the results, in a matter of business and technological outcomes.
Before going deeper on the results, a brief description might be needed of what OPENi is all about (or you just didn’t follow the link above). OPENi was an EU funded project in the area of cloud computing, and its goal was to create a personal cloud for each one of us, in order to become masters of our data and then choose who uses them and who doesn’t. In other words, everyone would have his own “personal Facebook“. It is still the ultimate dream for every European officer, in many cases citizen also, to break the chains from the “bad guy who sells our data for ads” and get independent. I am also in favor of such an utopia, we have to see how we can get there; and OPENi was an important lesson in that direction.
Then the difficulties started and the dream of a personal cloude was put into challenge. First of all, users do love their applications and the services they already use, and they won’t stop using them in favor of any new platform. Additionally, it was not possible for the project partners to both experiment on new staff and also build a competitor for every single service coming from Silicon Valley. For that reason, the first decision was to create an interoperable layer that connects the personal cloud with a set of the most popular services; thus users would have access on the services they own and they could orchestrate their data easily, until the personal cloud (aka “Cloudlet”) was quite mature to make them independent from any business solution. We indeed studied how this layer would me as scalable as possible, and how it would be easily extendable. We did work on what Facebook starter (and then stopped), and described what a Graph API should be; and we described it for other services outside social networking as well, like e-commerce or activity tracking applications. We followed API standards that started popping up, and we proposed a collaborative tool to allow developers build and maintain their APIs. This is what my lab and my team worked 100% on. But then the problems started: platforms started changing API versions, like Facebook, and even “destroy” endpoints with desired resources. Other services don’t like to give any API to developers, because they would challenge their business models; for example Amazon loves to rent infrastructure to other developers, but it hates to give any information to developers about their products, prices, shopping lists or purposes. It was a time that everyone wanted to build a platform and build a community of developers, and then suddenly they changed their mind (maybe Microsoft still tries to bring developers in Windows Phone 🙂 ).
But problems didn’t stop there. People in the Internet are spoiled with all these free services, they don’t understand that actually “they are the product” sold to advertisers. Of course their data individually have no meaning, but it is questionable if users would enjoy better or worse services if they paid. So, then it was the other challenge: how can we convince people to pay to host their data, even if they own them? The idea was to build on Raspberry Pi and let users run the Cloudlet as hardware whenever they like. But questions arose: who is willing to pay to buy a Cloudlet? How much technically savvy should they be to setup such an early-stage concept? How could they make it to have the available bandwidth to access information from their mobile devices, for data they have at their home under a common internet line? Who owns this device? How many should they have? Is it power-efficient? The answer to all these questions was that there is no-one who is not hacker or hobbyist that may be interested to run a server at home. Ask Owncloud or Diaspora how they feel about it. For all these reasons we decided to design a decentralized architecture; a service provider would host a farm of Cloudlets, without having any access right on users data, and people would pay for that service. Alternatively, we created a networking service, and the old business model “free for ads” would be enabled, with one big difference: there would be multiple players, and all of them would be operating under the same standards and protocols, thus they can be interconnected across different farms.
At the end, unless new technologies arrive, or a personal cloud is delivered together with the proper hardware (e.g. a hard disk with easy-to-use interfaces and high-end design that would be nice-to-have at home, the decentralized (and not the federated) approach seems more logical. But still someone has to convince the service provider to invest on such a project. Then another question arises: who trusts a telecommunication carrier or a government to become a service provider to host their personal data…?