2. MARCO METODOLÓGICO
3.5 Análisis estadístico en la consulta de datos
Interviewed: Gunnar Nilsson, Ericsson Date: 2016-06-27
Time: 09:00-10:00
Topic: Open Source within Ericsson
What does the current decision process for open source projects look like?
For simple small packages we have the following process in place:
All externally sourced software packages are made available trough the Bazaar, which is basically an e-commerce platform for software. Whenever R&D needs something that is not available in the bazaar, sourcing will do the acquisition and make the software available in the bazaar. For FOSS, there is a specific task force within sourcing called 3PPT. 3PPT evaluates all FOSS requests in terms of licenses and effects on existing IP.
How does Ericsson deal with FOSS contributions by employees/departments?
When R&D improves a FOSS package and wants to contribute this to the open source version, it has to be evaluated by the Open Source Core Team first. They look at the FOSS license, the global scale and the current (or future) patent licenses Ericsson has.
We are always very careful to contribute to copyleft-type licenses, as we have to determine the scope of the ‘virality’ of the copyleft. But for permissive licenses we usually don’t have any issues. Then we look at the global scale of the project. When it is a project with a small development team and not very wide adaption we have no objections, when the scale is larger we have to investigate more. Finally, we look at our patent portfolio, not only current licensing deals, but also future potential deals, and patents which we determine valuable. The combination of these three factor determines whether or not we allow contributions to the upstream project. But at the same time we are not looking to create our own internal forks of FOSS projects, we want to stay as close to the original project as possible.
Does this happen often and for what kind of projects?
FOSS General Directive R&D Bazaar 3PPT Open Source Core Team
We get FOSS contribution requests quite a lot, often just small stuff like bug fixes, or minor functions. But with functions the decision becomes more difficult, as we have to determine how the functionality of the software is changed, whether this is an innovation and how this relates to our patent portfolio.
We also get some requests to open source a project created by one employee or a department. This are usually just tools, nothing major. Usually this is open sourced because the centralized tools department notices the use of a non-authorized tool and has an alternative available. If this happens there will be no more budget to support the tool and the developer / department open sources the tool. They hope that the project will be picked up by some FOSS developer, but this is rarely the case. There is a misconception that FOSS developers are waiting for new projects to pick up, which doesn’t happen.
What do you think about the projects where Ericsson is a key contributor?
OPNFV, Open Stack, Open Daylight. These decisions become more strategic than tactical.
The current FOSS projects in the TelCo industry are being really hyped. Looking at the Gartner Hype cycle, these projects are probably somewhere close to the hype plateau and somewhere in the future they will experience this disillusion dip. I don’t believe in a fully automated, self calibrating network, each network will need specific integration and tuning. Even then there will always remain a human coordination element. While these open source projects are aiming to contribute to this, it will still take a long time for them to reach the stability and level of development needed. In the mean time there are already proprietary solutions available. Probably in the future a mix between proprietary and FOSS elements will be the optimal solution. These large operators have the idea that they themselves can be the ones integrating the FOSS projects into their networks and save costs by this. But looking at current examples, companies are usually hiring outside experts to integrate and maintain their FOSS projects, look at RedHat.
Even for more established projects such as Open Stack, the project is growing and growing because all of the contributors are selfish. Even though the project is mainly developed by corporations, they all take on their own area of functionality and work on this, expanding the functionality. Due to this process the boundaries of the project are getting further and further apart and maintaining and stabilizing the software becomes very difficult. When you look at the more successful projects, they are governed far stricter and the scope of the software is well-defined. This is currently not the case for the majority of TelCo FOSS projects.
How do OPNFV, OpenDayLight and OpenStack relate to Ericsson’s core business?
They are not the core business, we don’t have many patents in the area. While we have our own version Ericsson Cloud Manager, this is not a main income stream. It’s more like a side-stream, or a complementary. It’s nice to offer it as well.