Network Design Achieving Redundancy with PE Devices
Redundancy in service provider network infrastructure can be achieved by deploying two Provider Edge (PE) devices at a customer's site. PE devices typically function as routers, providing Layer 3 operations for on-site equipment and acting as a default gateway for customer networks.
The most common approach to redundancy is to implement BGP multi-homing at the customer edge between Customer Edge (CE) devices and PE routers. This configuration allows redundancy through BGP's routing protocol capabilities. Alternatively, Interior Gateway Protocols (IGPs) like OSPF or EIGRP can be used to load balance traffic to the service provider across the PEs.
While it is possible to use technologies like multi-chassis LAG (EtherChannel) to connect to PE routers, it is not ideal due to several considerations:
- PE devices are generally routers provided by the ISP, and EtherChannel is primarily used on switches. Even on router platforms supporting EtherChannel, coordinating with the ISP to set up their end appropriately can be challenging.
- Using a port channel does not facilitate active and backup PE configurations, nor does it allow running iBGP or FHRP Protocols such as Hot Standby Router Protocol (HSRP) or Virtual Router Redundancy Protocol (VRRP). Instead, high-availability options suitable for routers, like VSS or vPC, would be required, which are not optimal for PE devices part of an ISP's infrastructure.
While connecting via a Layer 2 port channel could eliminate the need for the ISP to support multi-chassis LAG, it requires ISP coordination. Typically, technologies like EtherChannel, MC-LAG, VRRP, and HSRP are more common within enterprise and data center networks rather than at the network edge.
The preferred solution for redundancy involves Layer 3 routing, which provides multiple path redundancy and allows PE routers to remain independent, reducing the need for ISP coordination.