How Cisco embraced a DevOps culture within its network engineering team

How did Cisco's network engineers eliminate snowflake configurations and human error? They started treating their infrastructure as code.

#1about 3 minutes

The challenge of scaling a traditional network engineering team

A globally dispersed, CLI-driven network team faced the need to scale its data center footprint and become more agile.

#2about 8 minutes

Identifying key pain points in manual network operations

Manual processes led to wasted skills on repetitive tasks, slow configuration changes, difficult troubleshooting, and a lack of change tracking.

#3about 4 minutes

Adopting infrastructure as code for network configurations

The team transitioned from device-specific configurations to machine-readable formats like YAML and JSON stored in GitHub as the single source of truth.

#4about 1 minute

Making time for transformation amid constant firefighting

The team had to strategically create time for process improvement despite being consumed by daily operational tasks and outages.

#5about 2 minutes

Shifting from a waterfall to an agile NetDevOps workflow

Inspired by SRE and DevOps teams, the network engineers adopted an agile methodology to replace their traditional box-by-box waterfall approach.

#6about 4 minutes

Applying GitOps principles to network automation

Using GitHub as the source of truth, configurations were treated like software code, deployed programmatically via APIs, and subjected to peer review and testing.

#7about 7 minutes

Maintaining a consistent desired state for the network

Automation was used to enforce a desired state, manage software versions consistently, and track ownership of network resources like VLANs and IP addresses.

#8about 4 minutes

How automation enables rapid infrastructure provisioning

A real-world example shows how automated deployment and QA checks reduced the time to provision over 50 hardware devices from hours to under 15 minutes.

#9about 3 minutes

Preventing undocumented changes or "snowflakes"

A disciplined, source-controlled process with full configuration replacement eliminates undocumented, one-off "hot fixes" that create long-term instability.

#10about 4 minutes

The outcome of a fully version-controlled infrastructure

The final result is a 100% version-controlled, auditable network that enables collaboration with other teams who can now contribute using machine-readable code.