Most IT managers are probably familiar with the concepts of DevOps and DevSecOps. There is a new model that is generating considerable excitement and is particularly relevant for networking professionals: NetDevOps.
As with other emerging technologies, the definition of NetDevOps varies depending on who provides it. But at a basic level, the term refers to the application of DevOps principles to computer networks.
“NetDevOps is a [and] trending topic,” said Andrew Lerner, research vice president for networks at Gartner. “But the first question is what is it, because there are different definitions and perceptions.”
What is NetDevOps?
According to Gartner’s definition, NetDevOps involves the application of DevOps concepts of continuous integration/continuous deployment (CI/CD) to networking activities. Other terms used to describe this model include NetOps 2.0, network as code, and GitOps networking, Lerner explains.
According to research firm GigaOm, NetDevOps “encompasses the abstraction, codification, and implementation of network infrastructure as code. [IaC]using programmed and automated workflows to eliminate configuration drift, thereby embedding quality and resiliency within the network.
According to Lerner, to take advantage of NetDevOps, an organization must have an automated pipeline that includes staging, pre/post validation, and testing of networking activities such as provisioning.
GigaOm agrees, adding that the NetDevOps pipeline can deliver code from different development environments to production, triggering comprehensive validation and conformance testing along the way.
NetDevOps also encompasses continuous monitoring, measurement and response, automatically triggering remediation alerts upon detecting configuration drift from a desired state, the company says.
What are the benefits of NetDevOps?
Given its DevOps roots, it makes sense that NetDevOps embraces many of the same goals. “DevOps is described by four principles: a holistic systems thinking approach (seeing the whole system, not just your part), no organizational silos, rapid feedback, and automation to reduce work,” says Joel King, an independent network automation architect.
Network operations have lagged behind other functional areas supporting an organization’s IT infrastructure, King says, especially when it comes to automation through network programmability.
“To deploy a new enterprise-enabled application or service, the compute, storage, and networking components all require configuration changes or the implementation of new hardware,” King says. “Often these changes are made manually, by an engineer typing on a terminal window, and may need to be reviewed by a formal change control process and implemented during a time slot that occurs outside of business hours.”
It can take weeks or months to make the necessary infrastructure changes to support application requirements, King says. “Automating infrastructure changes is fundamental to reducing the time it takes to deploy new applications or services to meet business needs,” he says.
NetDevOps “helps improve agility and is particularly valuable for organizations implementing infrastructure as code, since the network is often a bottleneck,” says Lerner. “NetDevOps practices generate clear workflows and documentation, which facilitates auditing, governance, and troubleshooting.”
Another benefit of NetDevOps is that it enables greater collaboration between different IT departments, just as DevOps does with development and operations.
“NetDevOps bridges the silos between network development and operations teams,” said Sameer Mahajan, principal architect at software company GS Lab. “These two teams have different responsibilities. The development team works on code development and testing, while the operations team takes care of the environment, IT architecture, deployment, availability, etc.
Traditionally, multiple transfers are involved when releasing code under development, Mahajan says. “This leads to software quality issues, delays, and frustration among team members,” he says. “NetDevOps solves these problems by automating this process and making it transparent. This increases the efficiency of deploying network changes and reduces errors. »
There are at least seven primary use cases for NetDevOps, Mahajan notes, including device provisioning, data collection and telemetry, compliance checks, migrations, reporting, troubleshooting, and configuration management. .
What are the obstacles to the deployment of NetDevOps?
To date, NetDevOps has not been widely implemented. “We estimate that only between 2% and 10% of companies are actively using NetDevOps practices today,” says Lerner.
Organizations must overcome a number of barriers to deploying NetDevOps:
- Risk aversion“The impact of a data center outage is huge,” says Lerner. “Thus, network teams are risk averse and lack confidence in automating data center networks. NetDevOps practices require an accurate repository of up-to-date network information (inventory, location, etc.), which is not common in many enterprises.
- Technical debt: Technical debt in the form of heterogeneous environments from a single vendor and configuration perspective is another hurdle, says Lerner. “And, of course, inconsistent or undocumented workflows related to network activities limit adoption,” he says.
- Skills shortage“We recommend investing in staff by redirecting hiring and training to specific software skills, including Ansible and Python, community forums, and cross-platform networking teams with adjacent DevOps staff,” says Lerner.
- Documentation: NetDevOps practices require an accurate repository of up-to-date network information, which is not common in many companies, says Lerner.
- Culture: For many companies, adopting these principles represents a major cultural shift, King adds. Organizations that have a low-trust, top-down management culture and have rigid change controls in place are likely to have problems implementing NetDevOps, he says.
How to get started with NetDevOps
Implementing NetDevOps is a matter of process, organization and training, rather than technology, says King. “A staff skills assessment is a mandatory first step,” he says. “Many network engineers don’t have the time, interest, or skills to become network programmers. However, some members of the network engineering team may find learning programming skills a new challenge and providing a learning opportunity will benefit the team.
Creating a team structure where subject matter experts (SMEs) can collaborate with programmers is a great approach, King says. Additionally, companies can create centers of excellence to demonstrate how an integrated team of SMBs and programmers works, and then apply and replicate that team approach to other areas of IT operations, he says.
In addition to investing in training, organizations preparing to move to NetDevOps need to give employees enough time to assimilate the new processes and tools, Lerner says.
“Train network teams with agile and DevOps teams,” says Lerner. He adds that network teams should capture and store both device configurations and network operational state (eg, active routing tables) in a version control system.
It’s also good practice to create standard templates for device types, enforce versioning, track configuration drift, and use automation where possible, Lerner says.
“Automate validation before and after changes. Automate pre- and post-environmental testing, such as latency/availability checks. And automate testing and configuration rollback,” he adds.
Ultimately, it’s about creating a new culture that supports NetDevOps, says Mahajan. “The organization should adopt the mindset of making small, frequent changes in an automated way,” he says. “Small changes make it less risky. Frequent changes help keep the environment as [up to date]. Automated mode reduces manual intervention and makes it less error-prone. »
How to maintain success on the road to NetDevOps
It’s one thing to start a NetDevOps initiative, but sustaining success over the long term also requires sustained effort.
“Sustaining a NetDevOps strategy requires management commitment to fostering and investing in time, training, and organizational structure to enable success,” King says. “These changes are long-term investments.”
It’s important to allow a team that can accomplish some small early successes, King said. “Capturing the parameters of these successes, in terms of cost savings, increased employee satisfaction and faster time to deliver value to stakeholders and make it visible to management, is critical,” he said.
Investing in continuous learning and experimentation by the team to expand their capabilities is also important, King says. “Give employees time for improvement work or innovation time, and adapt existing processes to accommodate those innovations,” he says.
How does an organization measure the success of NetDevOps?
“Success is measured by high network availability, fewer errors, and keeping the environment on the most recent version of the changes,” Mahajan says.
Businesses don’t have to go it alone when using NetDevOps. Although it’s more of a process than a commercial vendor offering, Lerner says, some vendors can help companies enable NetDevOps within their organization. These include Network to Code, Itential, HashiCorp, Arista, and others, which provide infrastructure as code and other network automation capabilities.
GigaOm says it anticipates a rapid evolution of the market in the coming months. “New players will emerge with lean and innovative solutions, while established network providers will compete by acquiring solution providers and developing critical partnerships.”
With different NetDevOps vendors and options available, and the changing landscape, selecting the best NetDevOps tools will depend on an organization’s use cases, existing software infrastructure, architectural choices, and internal capabilities.
For organizations looking to automate and accelerate the shift to digital operations, it’s likely that many will make NetDevOps the focus of their efforts to improve network performance and reliability.
Copyright © 2022 IDG Communications, Inc.