← Back to Projects

99.9% Uptime: Automated BGP Failover for Enterprise WAN

Architected a 99.9% uptime failover solution for premium enterprise customers using Multi-Homed BGP, automated via Route Maps and BFD.

“For premium enterprise customers, a single backhoe digging up a road meant a complete business outage. The ‘Last Mile’ was the weakest link in the chain.”

Executive Summary

In 2011, physical cable reliability in urban India was volatile. Road construction frequently severed primary fiber trunks, resulting in multi-hour business downtime.

I architected a Multi-Homed BGP failover topology combining high-speed Optical Fiber with an independent RF/WiMAX wireless link. This delivered guaranteed 99.9% uptime by enforcing automated, sub-second failover at the routing protocol level.

The Challenge

Physical infrastructure reliability was low, making fiber cuts an operational inevitability.

  • Single Point of Failure: Enterprise clients depended entirely on single-provider physical fiber drops.
  • Manual Failover Delays: Backup link cutovers required 30+ minutes of manual patch swapping by on-site teams.
  • Latency Sensitivity: VoIP and financial data streams required sub-second failover, rejecting the standard 3-minute BGP convergence timer.

[!IMPORTANT] Standard BGP keepalive timers (60s/180s) were far too slow for real-time applications. A 3-minute outage dropped every active SIP call across customer call centers.

The Solution

We implemented an Active/Standby BGP architecture with custom path manipulation and fast-failure detection.

Technology Stack

  • Edge Router: Cisco ASR 1002 Router (IOS-XE)
  • Routing Protocols: eBGP, OSPF, BFD (Bidirectional Forwarding Detection)
  • Transport: Metro Ethernet (Primary Fiber) + WiMAX RF (Backup Wireless)

Technical Architecture

1. Traffic Steering Policy

  • Outbound Traffic: Used Local Preference to force outgoing traffic to Fiber (Pref 200) over WiMAX (Pref 100).
  • Inbound Traffic: Applied AS Path Prepending to advertise the WiMAX path as three hops longer to upstream providers.

2. Sub-Second Failure Detection via BFD

We configured Bidirectional Forwarding Detection (BFD) to monitor link health at the hardware forwarding plane.

[!NOTE] BFD sends lightweight polling packets every 50ms. If 3 consecutive packets are missed (150ms), BFD signals BGP to immediately withdraw the primary route.

! # Define Policy for Primary Fiber Link
route-map RM_CUST_PRIMARY permit 10
 set local-preference 200
!
! # Define Policy for Backup WiMAX Link
route-map RM_CUST_BACKUP permit 10
 set local-preference 100
 set as-path prepend 65000 65000 65000
!
! # BGP Configuration with Sub-Second BFD Failover
router bgp 65000
 neighbor 10.1.1.2 remote-as 45184
 neighbor 10.1.1.2 route-map RM_CUST_PRIMARY in
 neighbor 10.1.1.2 fall-over bfd
 !
 neighbor 10.2.2.2 remote-as 45184
 neighbor 10.2.2.2 route-map RM_CUST_BACKUP in
 neighbor 10.2.2.2 fall-over bfd

The Outcome

  • High Availability: Maintained Zero Downtime across 3 major municipal fiber cuts during the 2011 monsoon season.
  • Bandwidth Optimization: Restricted expensive RF wireless spectrum usage strictly to failover events.
  • Revenue Impact: Introduced a “Gold Tier SLA” product line that commanded a 40% price premium.

The Verdict

Key Takeaway

Redundancy is an Illusion Without Fast Detection.

Having a secondary link is useless if your routing protocol takes minutes to react. Pair Multi-Homed BGP with BFD to ensure failover occurs before applications detect the drop.