TTR Networks

Beginner · Why not just keep it simple? · section 7 of 8

Simple, not simplistic

✓ Lab-tested 18 October 2026 on FRR 10.2.1

"Simple" and "simplistic" aren't the same thing. A simple network does what its job needs with nothing extra; a simplistic one leaves out things its job needs and hopes. These principles are how the lessons on this site try to stay on the right side of that line.

  1. Start with the simplest design that survives the failures you care about. Name the failures first, then choose the design.
  2. Every component must earn its place. If you can't say which failure, scale problem or outside network it handles, remove it.
  3. Redundancy you haven't tested is a hope. Fail things on purpose, in a lab and in maintenance windows, and time the recovery.
  4. Detect failures automatically, at both ends, quickly. Interface state isn't enough; protocols, keepalives and BFD are how routers find out.
  5. Make the common case boring. One way of doing each thing, the same on every router: templates, naming, numbering. Surprises are where outages hide.
  6. Know your normal. Prefix counts, session states, traffic levels. You can't spot "wrong" without knowing "right".
  7. Write down why. The next person (often you, a year later) needs the reason a piece exists before they can safely remove it.

Why each piece exists

Every technique on this site answers one of the three forces: failure, growth, or other networks.

We do thisBecauseLearn it in
Routing protocols instead of static routesFailures are detected at both ends, and the work doesn't grow with every subnetOSPF Fundamentals
BGP at the edgeOther networks only tell you what they reach through it, and you need policy, not shortest pathsBGP Fundamentals
BGP between loopbacksA session survives the failure of any one linkBGP on loopbacks vs interface addresses
BFDFailures detected in under a second instead of minutesBGP on loopbacks vs interface addresses
Two gateways with VRRPHosts keep working when a router failsOffice gateway: DHCP, DNS and VRRP on VyOS
Edge and core routersOutside-facing risk and internal routing change separatelyEdge and core
Route reflectorsiBGP sessions don't multiply as routers are addedRoute reflectors on VyOS
CommunitiesIntent travels with a route, instead of in someone's headBGP communities, prepending and RTBH
Filters, IRR and RPKIMistakes and attacks from other networks stop at your edgeRPKI and Route Origin Validation
A firewall in zonesPolicy stays understandable as interfaces are addedZone-based firewall on VyOS
Monitoring, backups, rollbackYou find out first, and can undoOperating VyOS