Wrap-up and resources

Key Takeaways

• 1. Monitoring the network: SNMP, logs and flows: Collect four things: state (SNMP), events (syslog with one clock from NTP), who talks to whom (flows), and second-by-second detail (streaming telemetry). Build a baseline before you set alert thresholds, and alert only on what needs a human.
• 2. Quality of service: voice first when the link is busy: QoS does not create capacity; it decides who waits when it runs out. Classify at the edge and mark with DSCP, put voice (EF) in a capped priority queue, give the rest fair shares, and shape traffic before the provider's link instead of letting the provider drop it.
• 3. Network automation: a source of truth, templates and checks before change: Good automation starts with data, not with a script: one source of truth, a template per device type, a diff before you push, a check after, and every change in Git reviewed by a colleague. Start with safe read-only tasks before you write.
• 4. Project: a branch with no single point of failure: A resilient network is not proven by a diagram but by testing failure: unplug every device and every link one at a time and time how long service breaks. What you do not test, you do not know works.

Glossary