New networking technologies are gaining traction in the industry as practitioners look for innovative ways to manage their complex environments.
However, misconceptions swirl around what new networking technologies can do for an organization and what they mean for network teams. For example, organizations consider using a digital twin for network management purposes as a worthy investment. In the same breath, many network professionals believe AI to be the catalyst that will set off a chain reaction of layoffs in the IT industry.
Three networking bloggers set the record straight on what these networking technologies can do for network management, and they identify approaches that can simplify the implementation process.
Identify risk vs. reward for digital twins in network management
It's feasible for a network pro to create a digital twin model of a network the same way an engineer would create a model of an IoT system. But, before doing so, practitioners must evaluate the risks and rewards of implementing a digital twin model for network management, wrote Tom Nolle, president of CIMI Corp., on his blog.
Digital twins would be more practical in software-defined networks that depend on explicit routing via a centralized controller, Nolle wrote. A digital twin could create an abstraction layer that represents different devices and elements in the environment. Digital twins would work better in networks that use static routing because they don't constantly change, as with adaptive routing.
Most networks use adaptive routers and switches, which means they adjust and interact with each other based on network behavior. These networks wouldn't benefit from a digital twin model because the model could interfere with how the routers readjust to the network. Disabling adaptive behavior in devices could eliminate this risk, but Nolle wrote there would be little to gain from configuring a model in these types of networks.
"Lack of risk doesn't constitute a benefit," Nolle wrote. "Can we identify anything interesting we could do with the digital twin model? Yes, but not much."
Nolle identified a few ways a digital twin model with abstraction could improve network management, such as the following:
- support software-defined networking (SDN) and adaptive routing;
- operate a network management system; or
- consolidate mixed router and virtual networks.
Despite these use cases, many vendors are reluctant to offer multivendor abstraction, largely because SDN hasn't advanced enough for them to offer these services. Likewise, teams are just as hesitant to deploy new technologies in their systems. Nolle wrote digital twin models could be difficult to put into practice, but network pros should "wait and see" how they might work in network management.
Network automation is less common than realized
All the buzz surrounding network automation might be no more than a few loud murmurs. Recent research from Gartner showed that adoption of network automation is less widespread than what the market implies. Over 50 network automation tools are available for enterprises, but automation only encompasses less than 35% of network activities, Andrew Lerner, vice president of research at Gartner, wrote in a blog.
Currently, few organizations automate more than half of their network activities. A clear divide exists between organizations with automated networks and those without. Enterprises with automated networks are more vocal in the industry, consequently creating a "false sense of widespread network automation," Lerner wrote. This results in vendors providing options to a small slice of the market rather than the majority.
In its "Market Guide for Network Automation Tools" report, Gartner outlined some obstacles preventing the adoption of network automation tools, which include budget constraints, limited skill sets and lack of confidence in using the tools. Gartner recommended organizations program easier "quick win" activities to get started with the automation process.
Some of those quick wins include the following:
- creating trouble tickets with network information;
- automating device configuration archives; and
- enabling or disabling monitoring tools when implementing a change.
AI helps rather than hinders
One of the biggest concerns about AI -- and one of the biggest reasons why AI adoption is limited -- is the notion that implementing the technology will lead to mass layoffs in the networking industry, as engineers lose their jobs to machines. However, the process of an engineer programming AI to automate network tasks suggests that AI will promote teams, Tom Hollingsworth, networking analyst at Foskett Services, wrote on his website.
AI would automate the trivial and mundane tasks of network operations, which are typically repeated tasks. As the burden of those chores offloads over to AI, network professionals would be free to focus on new or critical duties. This innovation, Hollingsworth noted, would enable AI to promote professionals into higher roles, as engineers can focus on complex tasks that AI can't perform.
"AI doesn't take away jobs. It takes away tasks," Hollingsworth wrote. "If your job is a collection of tasks that need to be done, then it's worth asking why it's so easy for it to be replaced by an AI system."
While AI will supersede some responsibilities, that doesn't mean network professionals will no longer be needed in those areas. A human will be the one to train and configure the AI algorithm, as well as update hardware or procedure changes.
"Given those constraints, AI will work for you, not against you," Hollingsworth wrote.