With growing number of "connected" devices, 5G technology promises to deliver accelerated connectivity to consumers along with lucrative services. But it has to complement this with automated internal operations which will speed up the introduction of new lucrative services delivered through agile networks. NFV and DevOps are the mainstream technology for many IT- and telecom-based companies. Organizations must refine their business strategies to meet speed and automation expectations along with flexibility, short time-to-market, and cost-effectiveness.
With NFV, network infrastructure has become software-centric by virtualizing core network functions (firewall, routing, DNS). The word software comes into the picture in NFV architecture to represent code written for developing virtual network functions (VNF) and a model describing infrastructure and execution environment for hosting VNFs. With software approach in NFV, telco and enterprise IT infrastructure using NFV can reduce the cost for network equipment, reduce power consumption, build a programmable centralized network, and reduce time-to-market for new initiatives for network consumers.
But still, challenges exist in the NFV approach for network service providers. Deployment and configuration of whole NFV infrastructure and making it completely available for delivering all NFV benefits is a complex and time-consuming process. The core of NFV architecture is the number of VNFs available and how VNFs are onboarded and aligned with workflows and operational lifecycle management within the infrastructure. To keep up the market competition, service providers want new services - VNFs - to be integrated with NFV infrastructure for consumers, and existing VNFs will need timely updates. A typical VNF integration has to go through three development cycles at three levels: solution provider, solution integrator, and a production environment that is NFV infrastructure. The development cycle includes development, testing, onboarding, packaging, and deploying to NFV infrastructure. This whole process requires manual efforts, long wait times for developers, and operators involved in pushing services to production, as well as backtracking efforts in case things don’t work as expected. To overcome this complexity in NFV, there is a need to reduce human intervention with automation of processes, removal of barriers in teams at VNF providers' and service providers' end, and incremental deployment of VNFs.
Here, DevOps comes into the picture. DevOps is a new culture in software engineering wherein software development and delivery become automated and collaboration and communication are established among developers and network infrastructure operators. DevOps provides a defined set of methods, tools, and platforms for programming and execution of services. It includes easy to use SDKs, agile development platforms, test automation tools, and reusable libraries. With such benefits from DevOps, converging DevOps with NFV service providers reduces the manual effort of VNF onboarding and infrastructure configuration by automating processes and decreases the time-to-market for new services.
When we align DevOps for NFV, we can
- Automatically deploy and test new updated network services.
- Perform testing and benchmarking of NFV infrastructure components.
- Achieve performance and reliability policy-based, fast and resource-efficient VNF onboarding, deployment, scaling, and resiliency.
- Automatically tune VNF deployment parameters like the number of CPUs and RAM during deployment or runtime.
- Achieve real-time checking of expected VNF operational behavior in NFV infrastructure.
- Automatically scale and patch resources with real-time monitoring events.
The DevOps approach enables NFV architecture to utilize Intent Based Networking (IBN) in the NFV ecosystem. Gartner researcher Andrew Lerner wrote on their blog about how IBN is going to transition networks. In January 2018, Cisco launched an IBN-based solution to strengthen the IBN market. IBN is kind of a subset of the machine learning concept which allows network admins to convey instructions on what operation needs to be done and what output is expected. This way, IBN takes DevOps-enabled NFV infrastructure to the next level by translating instructions into action steps and further automating lifecycle management of VNFs and NFV components.
Currently, for 5G networks, NFV is going through the development phase, where several solution providers will overcome complexities using existing technologies. However, service providers and network administrators should keep in mind the future use of machine learning capabilities enabled with Intent Based Networking. Having an agile and automated NFV model will set the stage for such luxuries for service providers. We can still make an IBN case for smaller NFV deployments due to the criticality of operations.
DevOps Tools for VNF Onboarding
There is a range of possible tools available to use for implementing DevOps for VNF onboarding in NFV infrastructure, but CSPs need to decide which are right for their environment.
- Code Management/Version Control: Git, Gitlab
- Issue Management: Jira, Maven
- Test Automation: Selenium
- Binary Management: jFrog Artifactory, Docker Registry
- Scalability logging: vElite Ek
- Monitoring & Automated Recovery: Wily, Dynatrace
- Configuration management tools: Puppet, Salt, Ansible, Chef
- Continuous Integration: Jenkins
- Flexible system deployment: VagrantS
The role of NFV architecture in 5G deployment is extremely crucial, but the overall deployment and transition to NFV are quite challenging in terms of added complexities at the operational level and infrastructure deployment level. With NFV, network functions and operations become programmable, meaning more APIs are enabled, and can easily bring them into the DevOps chain. Applying agile software development practices and DevOps to an NFV infrastructure can help service providers to create new services faster and more dynamically to gain a competitive edge and generate revenue.