Open source software is one of the key building blocks of data centers. It is a vital enabler for the transition to open ecosystems as carriers modernize their networks. But many struggle to get their arms around the real value of open source and if its “free” brand is the key to reducing systems cost.
In order to describe the business dynamics around open source-based platforms, I have found analogies with ice hotels supportive. Ice hotels differ from classic hotels in three main ways. The building material, ice, is free. The hotel opens as soon as the reception and the first few rooms are completed, and the ice hotel then evolves while in operation. And the assets have a short life span, which makes life cycle management very different from a classic hotel. It might have been this analogy the governors of OpenStack thought of when they named the ninth release “Icehouse”.
Open source-based platforms build on free components. The full value of the components are realized when they are integrated together into a platform or solution and when support can be provided for the larger entity. Free components do not mean the broader solution is free. The pace in the development of components is very high, with 400 new features in the Kilo release of OpenStack alone. Platform and solution providers strive to develop a stable foundation and a dynamic ecosystem based on these free components. In other words, just because the ice is free doesn’t mean the cost of building and supporting an ice hotel is zero.
Open source and a dynamic ecosystem are tightly coupled to continuous introduction of new capabilities in the system. New features and capabilities are added to systems in operation, and development and operations are closely coupled. For instance, the DevOps methodology blurs the boundaries between development and operations, and new software components can be introduced in networks weekly or monthly instead of yearly. These also create a new conundrum with regards to what should remain industrialized and tested software components and what should be developed as custom-designed parts. Ice hotels are also in continuous development. They open early. New rooms are added throughout the season. Then the platform can become unstable at the end of the season, calling for a major system upgrade.
Life cycle management in an open source world represents a similar balancing act. The platform evolves with the underlying software releases, for example, every 6 months with OpenStack. Features migrate from custom components into the open source base. Solutions leveraging the platform evolve rapidly, and the migration between releases is faster than in a classic network. The high level of continuous integration activities is a vital part of the life cycle management cost and the economic life span for each release is significantly shorter than for classic network deployments.
My predictions for the introduction of open source software in the network are:
- The primary value of an open source ecosystem is agile feature introduction and broad industry support.
- The transition to a DevOps model has the potential to create a software platform with 10x shorter cycles between updates, compared to classic IT and network releases.
- The total capex for an open source platform is driven by the integration efforts and support of “free” software.
- Finding a good balance between industrialized blocks and customized components determines overall total cost of ownership (TCO).
- The operational set-up for open source-based platform and ecosystems is very different from what we are used to in classic broadband and mobile networking.