It’s not something you want to spend a lot of time thinking about—after all, network automation isn’t a topic you bring up at parties or on a first date. But you probably think about it, a lot. Because while automation offers the promise of simplicity, there’s no simple path to get there.
Throw artificial intelligence and advanced analytics into the mix, and simplicity just got a whole lot more complex.
These days, when it comes to network automation, it’s not just about simplifying manual, repetitive tasks with one-off scripts. And a single off-the-shelf solution is rare, says Shamus McGillicuddy, research director with Enterprise Management Associates, in a webinar about EMA’s report, Enterprise Network Automation for 2020 and Beyond.
In EMA’s survey of 250 IT professionals, most say they’re using multiple tools for automation (40 per cent use two tools, 34 per cent use three and 16 per cent use four or more). The most popular option was third-party, multi-vendor network automation software, followed closely by general IT automation software adopted to network services.
The way networking pros traditionally handled automation — one-off scripts for single tasks — isn’t going to cut it going forward. Network environments increasingly span multiple domains and multiple clouds, so automation is more necessary than ever.
“Networking pros can’t be mired in manual, time-consuming networking configuration and command-line tasks. They need to be able to bring change and flexibility through software-based programming that is automated, less error-prone and repeatable,” according to a Cisco blog from Cisco Live earlier this year.
The network can also be used for gleaning data insights, thanks to advanced analytics or AI/Ops, which has become an important element of automation. Machine learning and statistical analysis can “inform your use of automation,” says McGillicuddy. In other words, it can “tell you what you should do with your automation tools.”
So, what exactly can you do? The top use case, according to EMA’s report, is driving automated change from analytically derived insights. Other use cases cited by respondents in the survey include:
- optimizing network health and performance
- recommending automated actions to a human operator
- translating business logic into network actions
On the data centre side, network security and anomaly detection are top automation priorities, followed by virtual networks. “A lot of people have a lot of manual management around those overlays,” says McGillicuddy, such as configuring micro-segmentation.
On the enterprise side, automation priorities often involve wireless LANs, public cloud (IaaS) and WAN edge/branch routing, including SD-WAN, according to EMA.
But wait? Doesn’t SD-WAN imply some level of automation? Well, yes, but “the depth of automation is shallow,” says McGillicuddy. “For the most part … they’re only automating less than a third of tasks.”
Newly released research from Juniper Networks in its State of Network Automation Report, or SoNAR, found that—while it may seem counter-intuitive—focused efforts yield better results than automating across the entire network.
“It’s imperative for networking professionals to focus their efforts and resources while automating more deeply across a smaller cross-section of the infrastructure to transform culture, skills, processes and a toolset that can adapt to any place in the network once refined in a focused space,” according to the report.
The end goal of network automation is, in part, simplification. But it’s not easy getting there. On the technical side, network automation tools can be difficult to build or implement, and they doesn’t always scale. Even off-the-shelf solutions require customization.
On the business side, it’s expensive and requires a cultural shift. Staff may distrust automation and refuse to use it, says McGillicuddy. There might also be push back from the “old guard” who are worried about the prospect of automation taking away jobs.
Automation holds a lot of promise for IT networks, but it’s best not to jump in without a plan. Start small, develop the required skill sets among your IT team and focus your automation efforts. Because the road to simplicity is paved with complexity.