(Documentation, Requirements, Integration & Testing, Collaborative Development)
A well established test bed is a critical resource to achieve opnfv's goals. Among these goals, integration of upstream source distributions, operational tools, other software components, hardware infrastructure, validation tools, well thought out quality assurance, well thought out performance metrics benchmarking,… rely heavily on a well established / well understood test bed.
Many developers may not have either the resources or the time or the inclination to create and maintain such a test bed on their own. It is a substantial challenge to create such an infrastructure in sizable / realistic scale. The well designed / well documented "sandbox" environment can improve developer efficiency and accelerate opnfv's project progress. This benefit can be extended most of the projects.
Benchmarking a network function can be a complex undertaking with many variables. This project focuses on a bottom-up approach and will likely meet the need of another project that focuses on specific ETSI defined use cases from top down.
Test bed infrastructure and tooling project extends to all the SW and HW components needed to establish and operate such a test bed. It should:
The test bed will support the following functionality:
In high level, the interfaces are:
For the first release, may be only the "user interface" is required.
Release candidates are delivered from CI project. We need to work out a release candidate cadence in practice to (i) have stability of the test bed for developers who rely on a stable environment, and (ii) meet the need of testing the latest release candidate. A suitable cadence needs to be worked out, or a rolling scheme.
More complex NFV use cases may be built on the top of this test bed platform. Those use cases may have separate projects / subgroups dedicated to each of them. Additional tooling may be required for specific use cases. But commonly shared test tooling should all be included in this project.
Some of the testing and benchmarking may be significantly expedited or improved if we can leverage commercial products, either SW and HW. In these cases, some software tools may be required to make the resource available for shared use.
Debugging and tracing are part of main deliverable for all in scope testing/benchmarking.
Test plans of platform level and tool level instrumentation will be covered in this project. It is assumed that a sister NFV use case project(s) will produce higher level use cases in a top-down approach.
Covered in this project includes tools that may go to a developer's client, e.g. for emulation of certain network connectivity. Most of other "client" tools may be supported with web UI.
See above in "Scope" section.
Out of scope:
The project may be expanded in several dimensions: Just some examples.
This will be part of the first release.