How to deploy redundant BNGs

netElastic’s BNG supports N+1 redundancy in either active/active or active/backup modes.

To achieve this, both BNGs should see all PPPoE or IPoE requests from the L2 access network. Both BNGs should route to the upstream core routers using a dynamic routing protocol such as OSPF or BGP to allow subscribers to be routed out by either BNG.

The BNGs will be configured to automatically load balance the incoming requests or act as primary/secondary. This is accomplished using PADO delay settings for PPPoE, or DHCP reply delay for IPoE.

The PADO or DHCP delay can be controlled at the MAC address level, dividing up odd and even MAC addresses for BNG1 vs BNG2. (e.g., BNG1 having no delay for odd MAC and short delay for even MAC. Visa-versa for BNG2.)

Example configuration – BNG1

vci-configuration
 interface eth-trunk2.10
   pppoe template my_PPPoE_template
   max-ipox-session 0
   max-pppox-session 10000
   encapsulation ppp-over-ethernet
   access-delay 500 mod-mac 2 remainder 1
   ip-access-type dual
  exit
exit

Example configuration – BNG2

vci-configuration
 interface eth-trunk2.10
  pppoe template my_PPPoE_template
  max-ipox-session 0
  max-pppox-session 10000
  encapsulation ppp-over-ethernet
  access-delay 500 mod-mac 2 remainder 0
  ip-access-type dual
 exit
exit