What new demands will our network face in 2025? In this blog series, the Cisco IT networking team will share our vision for the future of our network—and the investments we’re making to get there.
In collaboration with Dipesh Patel
Insights and automation will power our future network. Think of it as a circular process: collect data from network infrastructure. Analyze it for insights. Share those insights with teams to help them improve service. Use the insights to automatically reprogram infrastructure where possible. Repeat. The aim is to quickly adapt to whatever the future brings—including new traffic patterns, new user habits, and new security threats.
Now I’ll dive into more detail on each block in the diagram.
Data foundation. Good insights can only happen with good data. We collect four types of data:
- Inventory data for compliance reporting and lifecycle management
- Configuration data for audits and to find out about configuration “drift”
- Operational data for network service health monitoring
- Threat data to see what parts of our infrastructure might be under attack—e.g., a DDoS attack on the DMZ, or a botnet attack on an authentication server
Today, some network data is duplicated, missing (e.g., who authorized a change), or irrelevant. To prepare for our future network, we’re working to improve data quality and store it in centralized repositories such as our configuration management database.
Analytics. With a trusted data foundation, we’ll be able to convert data to actionable insights. We’re starting by visualizing data—think color-coded dials—to make it easier to track key performance indicators (KPIs) and spot trends. Examples of what we track include latency and jitter for home VPN users, and bandwidth and capacity for hybrid cloud connections. We’re also investing in analytics for decision support. One plan is tracking the number of support tickets for different services so we can prioritize the work with the biggest impact. Another is monitoring load and capacity on our DNS infrastructure so that we can automatically scale up or down in different regions based on demand. Currently, we respond to performance issues manually—for instance, by re-routing traffic to avoid congestion. In our future network we’ll automate changes in response to analytics. Which leads me to our next topic: automation.
Policy and orchestration. February 2022 marked a turning point: we now fulfill more change requests via automation than we do manually. As shown in the figure, we automatically fulfilled more than 7,500 change requests in May 2022, up from fewer than 5,000 just six months earlier. Examples include automated OS upgrades with Cisco DNA Center Software Image Management (SWIM), compliance audits with an internally developed tool, and daily configuration audits with an internal tool we’re about to swap out for Cisco Network Services Orchestrator. We have strong incentives to automate more and more tasks. Manual activities slow things down, and there’s also the risk that a typo or overlooked step will affect performance or security.
In our future network, automation will make infrastructure changes faster and more accurate. Our ultimate goal is a hands-off, AIOps approach. We’re building the foundation today with an orchestrator that can coordinate top-level business processes and drive change into all our domains. We are working closely with the Cisco Customer Experience (CX) group to deploy Business Process Automation solution. We’re developing workflows that save time for staff by automating pre- and post-validation and configuration management. The workflows integrate with IT Service Management, helping us make sure that change requests comply with Cisco IT policy.
Release management. In the past, when someone submitted a change request one or more people manually validated that the change complied with policy and then tested the new configuration before putting it into production. This takes time, and errors can affect performance or security. Now we’re moving to automated release pipelines based on modern software development principles. We’re treating infrastructure as code (IaC), pulling device configurations from a single source of truth. We’ve already automated access control list (ACL) management and configuration audits. When someone submits a change to the source of truth (typically Git), the pipeline automatically checks for policy compliance and performs tests before handing off the change for deployment.
The Road Ahead
To sum up, in our future network, the only road to production is through an automated pipeline. Automation helps us adapt more quickly to unexpected change, keeps network configuration consistent worldwide, and reduces the risk of errors. We can’t anticipate what changes our business will face between now and 2025—but with insights and automation, we’ll be able to adapt quickly.
Learn more about our strategy: