Network automation technology must understand the infrastructure that it is acting upon. This understanding requires comprehensive data on the configuration of a network and the state of that network. Increasingly, industry leaders refer to this comprehensive data repository as a network “source of truth.”
New research by Enterprise Management Associates (EMA), “Enterprise Network Automaton for 2020 and Beyond,” found that 98 percent of enterprise network automation initiatives have some flavor of a network source of truth, and 41 percent of enterprises consider their source of truth as essential to automation. This research, which was based on a survey of 250 IT professionals directly involved in a formal network automation initiative, found that enterprises that are successful with network automation are more likely to call a source of truth essential.
As a network reliability engineer with a midsized global media and entertainment company told EMA, “Our [network automation] is driven by a central database that we call a source of truth, where we put in network intent. Then, all our automation tools are integrated around that database, and our tools generate configs based on that information and deploy it. We have all our monitoring and alerting tools hooked into that database to compare every alert coming in and understand what was worth alerting on.”
The Core Contradiction: A Single Source of Truth is Rare
A network source of truth will contain a variety of data, such as configurations, device state metrics, IP addresses, device inventory information, policy rules, and more. These classes of data rarely coexist in the same database.
When network automation industry leaders talk about a source of truth, they are usually talking about multiple repositories of information. In fact, only 26 percent of surveyed enterprises claimed they had a single network source of truth for automation. Instead, 63 percent have multiple authoritative data repositories for different classes of information.
“There isn’t a single source of truth” a network automation engineer with a large North American entertainment company told EMA. “There are systems of record, like IPAM for addresses, data center intrastation management software that has a record of devices on the network, and another that tracks cabling. But it’s always ephemeral.
“We ended up with multiple repositories for our source of truth,” the network reliability engineer with a midsized global media and entertainment company told EMA. We used Netbox as the primary database, but there were a lot of things we couldn’t store in it. So, we built an additional source of truth in Git and merged information from different places.”
How Does a Source of Truth Help with Automation?
EMA asked enterprises to reveal how a source of truth best supports their network automation initiatives. Improved integration or compatibility with other IT systems was the most important to them, as selected by 35 percent. With an accurate understanding of the network, automation tools are better able to integrate with IT service management, IT orchestration, SDN controllers, and other technologies.
The second top benefit of a network source of truth was reduced security risk (34 percent). When engineers and administrators have an authoritative data set on network configuration and state, they can push change through an automation pipeline with more confidence that changes won’t introduce inadvertent security vulnerabilities.
Finally, 29 percent of enterprises said a source of truth helps network automation tools to rapidly support new network features and technologies, and 24 percent said a source of truth promotes network reliability.
The network reliability engineer with a midsized global media and entertainment enterprise told EMA, “Consistency is the biggest benefit because we have this database that represents the network, and it allows us to deploy things in a consistent way that is defined by how we want the network to be.”
EMA believes that enterprises must research the concept of a source of truth vigorously before the embark on a network automation initiative. IT leaders must understand what kinds of data will be important for their own source of truth. They must explore how the network automation solutions on their technology shortlists consumes this information to drive effective and reliable automation on the network.