EMA research found that enterprises use network analytics technology to automate a variety of networking tasks for increased uptime and other benefits.
Network analytics is a hot topic these days. New tech startups wear the “network analytics” label with pride, and suppliers of network operations software and network infrastructure are investing heavily in analytics to enhance the value of their products. While there are countless reasons for a surge in analytics interest, I believe network automation is one of the biggest drivers.
Two years ago, my research found that half of enterprise network management teams had an active initiative to apply advanced analytics to network data. That finding prompted me to launch new research this year on the topic. I surveyed 200 IT professionals who were directly involved in a network analytics initiative, and published my findings in the Enterprise Management Associates (EMA) report, “Network Analytics: Applying Machine Learning and More to Network Engineering and Operations.” (You can find a recording of a free webinar that highlights the results here).
The majority of these initiatives consumed network analytics in three ways. First, they used analytics features embedded in network infrastructure. Second, they used analytics features embedded in network operations software. Third, they bought products from network analytics specialist vendors.
Analytics and automation are interconnected
My research found that network analytics is intrinsic to network automation. When we asked respondents to identify the broad, technical initiatives driving their interest in network analytics, IT automation was their No. 1 response by a significant margin (34% of all respondents). Regardless of drivers, however, automation is tied to nearly every network analytics project we surveyed. Ninety-seven percent of respondents said their analytics projects are supporting or enabling network automation today or will support it in the future.
Network automation has been around for a while. For decades, network engineers have been writing custom scripts to automate common, easily repeatable tasks in a network device’s command-line interface (CLI). First-generation network automation tools like network change and configuration management systems are good at automating iterative tasks, like pushing a simple configuration change to multiple switches or routers. They can also provide change management controls by comparing changes to network policy.
However, network automation has typically failed to support complex and unique networking tasks due to lack of insight. It's difficult for automation tools to understand two things: the intended state of the network and the actual state of the network. This is where analytics plays a key role.
The next wave of automation solutions, such as intent-based networking, include advanced analytics algorithms that can discover, translate, and draw comparisons between network state and network intent. They can learn how a network manager wants the network to run and monitor the network to validate that it is running as intended. However, this use case is just the starting point.
EMA asked enterprises to identify the types of tasks they are automating with support from network analytics technology. Network optimization (56%) topped the list. This goes beyond translating network intent to network state. Instead, these enterprises are trying to find the best network state. Route analytics can automatically steer traffic through the ideal path on the network. Other algorithms can dynamically tune QoS settings to protect application performance and user experience. For example, the CEO of a company starts using a new video chat application. A good analytics technology would prioritize this application, especially if the other end of the chat is the company’s board of directors.
A majority of network analytics initiatives (51%) were also automating cost optimization, which can manifest in a number of ways. Many enterprises deploy third-party network services in IaaS clouds, for instance. Every load balancer and firewall that is deployed with a cloud workload has a license associated with it. Those licenses can add up to big dollars over time. Many times, cloud admins will request these network services for a workload, and the network team will assign licenses to them. But those admins neglect to retire the network services when the workload retires. A good network analytics solution can track when those services are no longer needed, automatically retire them, and return those licenses to the pool.
Many of the enterprises surveyed reported automating network diagnostics and troubleshooting (43%) and change management (42%) with support from analytics. A smaller number (36%) go even further, automating fault and performance remediation, as in closed-loop operations.
The bottom line
We asked all these enterprises what kinds of benefits they were experiencing from network analytics. Among those whose analytics initiatives were specifically driven by an IT automation initiative, 79% had accelerated their ability to repair network problems. Sixty percent were able to improve their compliance with industry requirements and regulations. Finally, 46% were able to deploy new IT services more quickly. Overall, when network teams combine analytics with automaton, they increase uptime, reduce compliance risks, and accelerate service delivery. What IT organization wouldn’t love to experience those benefits?
EMA recommends that IT professionals keep in mind this link between automation and analytics. When vendors talk to you about their network analytics solutions, ask them about how analytics can help you automate your network. And when vendors approach you with a network automation solution, ask them how they are using analytics to make that solution more valuable and reliable.