OpenStack and NFV—agile networks and open source code
During the past decade, open source development has become more and more relevant—regardless of which part of the high-tech industry is scrutinized—and that has resulted in more open source components being added to both enterprise IT and telecom products. In parallel with this tendency, the emerging presence of network functions virtualization (NFV) and software-defined networking (SDN) has been moving the industry towards network and application virtualization. Because the latest trends in virtualization pushed several parts of the IT world onto cloud-based platforms, following this pattern became a logical move and next step for telecom applications as well.
OpenStack was started as an IT-focused initiative nearly six years ago and since then has become an open source de facto cloud platform. Today it serves as the basis of several private and public cloud deployments around the globe, and this is still only the start of the journey.
More and more telecom vendors are looking into adding new functionality to OpenStack to fulfill the needs of virtualized network functions (VNFs) running on top of a carrier-grade cloud platform, and it is still a challenge to synchronize the requirements coming from both enterprises and telcos to create a common road map.
Open Platform for NFV (OPNFV) is a carrier-grade, integrated, open source platform for accelerating the introduction of new NFV products and services. OPNFV is an open source project that was established one-and-a-half years ago, and Ericsson was one of the founding members. Since its launch, OPNFV has produced two releases called Arno and Brahmaputra, which are the realization of the NFV Infrastructure (NFVI) and Virtualized Infrastructure Manager (VIM) components of the ETSI NFV Architectural Framework, in which OpenStack is the main building block among elements such as KVM, Libvirt, OpenDaylight, and so forth.
Because the parts are continuously changing, this is a complex puzzle to put together. As a mid-stream project, OPNFV not only consumes the stable releases of the upstream projects but also participates in the development activities and contributes features, documentation, and bug fixes.
For instance, OPNFV Doctor is targeting the fault management area and focusing on two use cases: host failures and maintenance. The Doctor team members, which include developers from Ericsson, have been actively contributing to OpenStack since the Liberty release cycle; improving the alarm functionality, notifications, and reaction times in OpenStack; and building a large ecosystem out of the OpenStack and OPNFV communities by showing a good example of collaboration.
The first two modules where we landed code were Aodh—which is the new alarm module extracted out from Ceilometer—and Nova, where features that are critical for telecom applications and services were added. Because it is very important to improve the information flow between OpenStack and the applications on top, Aodh now supports raising alarms based on events emitted by OpenStack services, such as notifications about the state changes of virtual machine instances.
If you would like to try the new functionality, you can do that by downloading and installing the second release of OPNFV, called Brahmaputra, which contains Aodh as part of the reference platform. The team is continuing the work in Nova and Aodh, and we are also working in other modules to improve failure detection time, add new notifications, and provide advanced alarm management.
I have talked much about features and contributions, but that is not the only activity for OPNFV. We all know how important testing is – we can never have enough test cases and test runs to fully prove something is working and is doing what it was designed to do. The OpenStack team is continuously working to improve the continuous integration (CI) system and increase test coverage, but the team still doesn’t know how OpenStack works when it is part of a larger system. This is where OPNFV comes into the picture by providing functional testing, verification, and validation of the platform and infrastructure.
Similar to other open source projects, OPNFV is building a CI and continuous deployment (CD) pipeline and continuously enhancing its frameworks to have a fully automated test environment. By establishing these environments, the OPNFV labs can be used to test functionality and performance for different configurations of the platform, which are called scenarios in OPNFV.
This also means you can test OpenStack in slightly different setups and provide feedback based on the results. If you would like to learn more about this area, you can check the activities of the test projects, for instance Yardstick and Functest, or visit the RelEng wiki page to get more information about tools.
Ericsson is an active member of both projects, and these are only the first steps of a long journey to bring different parts of the industry closer by building on and further enhancing the same foundation.
Thank you for reading. Enjoy the Summit!