The Open Platform for NFV project has unveiled its fourth release, named Danube, that aims to work better with other open source projects and improve NFV testing.
Releases such as these are not for the faint-hearted, aiming as the do to address the almost infinitely complex software challenges and interdependencies required to virtualize network functions. The headline ingredients of this one include a dollop of improved testing at dash of functional support for MANO, and a sprinkling of DevOps magic.
“Danube represents an evolutionary turning point for OPNFV,” said Heather Kirksey, director at OPNFV. “It brings together full next-gen networking stacks in an open, collaborative environment. By harnessing work with upstream communities into an open, iterative testing and deployment domain, we’re delivering the capabilities that truly enable NFV, and that is very powerful.”
Here are those key ingredients as described in the OPNFV announcement.
Key enhancements available in OPNFV Danube include:
- Foundational support and introduction of capabilities for MANO: Integration between NFV Infrastructure/Virtual Infrastructure Manager (NFVI/VIM) with Open-Orchestration (Open-O) platform (now ONAP); instrumentation of NFVI network telemetry to support Service Assurance and other use cases; multi-domain template support (Domino project); and translation features between YANG and Tosca modeling languages (Parser project).
- Enhanced DevOps automation and testing methodologies bring a fully integrated CI/CD pipeline, the creation of Lab-as-a-Service (LaaS) to enable dynamic provisioning of lab resources, the introduction of stress testing into the OPNFV test suite, and a Common Dashboard that provides a consistent view of the testing ecosystem.
- Focus on NFV performance including acceleration of the data plane via FD.io integration for all Layer 2 and Layer 3 forwarding (FastDataStacks project), and continued enhancements to OVS-DPDK and KVM. The release also sees a renewed focus on performance test project activities through virtual switch testing (VSPERF project), root cause analysis for platform performance issues (Bottlenecks projects), initial compute subsystem performance testing to lay the groundwork for Benchmarking As a Service (QTIP project), and storage subsystem performance testing (Storperf project).
- Key NFV architectural enhancements, including the ability to dynamically enable and configure network control through integration with OpenStack Gluon and increased reliability and test cases that support multi-site and High Availability (HA) work.
- Feature enrichment and hardening in core NFVI/VIM functionality such as IPv6, Service Function Chaining (SFC), L2 and L3 Virtual Private Network (VPN), fault management and analysis, and a continued commitment to support multiple hardware architectures, as well as traditional hardware OEMs, whitebox, and open source hardware through collaboration with the Open Compute Project.
OPNFV’s Tapio Tallgren has also blogged on the matter and you can find further analysis at Light Reading. As the summary above illustrates, NFV is incredibly complicated, which is one of the main reasons it’s taking so long to come into effect. Clearly development needs to happen in an open and collaborative environment, so it’s good to see regular progress from OPNFV.