Hierarchical VPLS (H VPLS) has been scaled up from “flat” VPLS to satisfy the needs of big service provider networks. At least n(n-1)/2 connections are needed to connect every Provider Edge (PE) router in a flat VPLS to form a full mesh of pseudowires (PWs). The necessary signaling and replication overhead increases exponentially with the number of PEs (n), rendering big deployments unfeasible.
The H-VPLS Architecture
Scalability problems are resolved by H-VPLS, which divides the network into a two-tier hierarchy made up of two main parts:
- User-facing PE (U-PE): Smaller edge devices that link straight to the Customer Edge (CE) are known as User-facing PEs (U-PEs), sometimes called Multi-Tenant Units (MTUs). Instead of keeping a complete mesh, a U-PE uses a Spoke Pseudowire to link to one or two “Hub” routers.
- Network-facing PE (N-PE): Often referred to as PE-rs (Routing and Switching), network-facing PEs (N-PEs) are core routers that combine traffic from several U-PEs. Similar to a flat VPLS, N-PE devices maintain a Full Mesh of pseudowires among themselves.
You can also read How Data Flows Through The OSI Model And It’s Importance
How H-VPLS Functions
A U-PE, which carries out common bridging tasks including MAC address learning and frame replication, is where traffic enters the network. The U-PE uses a point-to-point spoke pseudowire to transmit this communication to an N-PE. Then, the N-PE routes the traffic to other locally associated U-PEs or other N-PEs in the mesh via the high-speed MPLS core.
In H-VPLS, two key techniques are established for the access domain:
- H-VPLS with Q-in-Q: This adds an extra S-VLAN tag to tunnel customer traffic when the U-PE is an Ethernet switch.
- H-VPLS with MPLS: When the access layer signals pseudowires between the U-PE and N-PE using MPLS labels, it is known as H-VPLS with MPLS.
Loop Prevention: The Split-Horizon Rule
H-VPLS uses a particular variant of the split-horizon rule to enable traffic to move across the hierarchy levels while preventing forwarding loops:
- Mesh to Mesh: A core mesh pseudowire’s traffic is never sent to another mesh pseudowire.
- Spoke to Mesh (and vice-versa): Traffic between spoke pseudowires and the core mesh can be freely passed from spoke to mesh (and vice versa).
- Spoke to Spoke: If two spokes are part of the same VPLS instance, their traffic can be routed to each other on an N-PE, enabling the N-PE to function as a switch for the U-PEs it is linked to.
You can also read What is Application server and How Application Server Works
Scalability Benefits
H-VPLS offers a number of technical benefits over flat VPLS, including:
- Decreased Signalling: U-PEs only keep a small number of spoke connections, which significantly lowers the number of active pseudowires in the network and causes signaling growth to be linear rather than exponential.
- Effective Replication: In flat VPLS, each PE in the mesh requires that the ingress PE replicate Broadcast, Unknown Unicast, or Multicast (BUM) packets. In H-VPLS, core distribution is handled by the N-PE after receiving a single copy from the U-PE.
- Optimised Hardware: Hardware Optimization Access devices (U-PEs) only need to learn local MAC addresses and those reachable via their particular N-PE, not the entire network, therefore they don’t need large MAC tables. This enables the employment of less complex and costly edge devices.
Redundancy and Fault Tolerance
Dual-homing is used to prevent a single point of failure, in which a U-PE loses connectivity in the event that his principal N-PE or speaking PW fails. It is possible to configure a U-PE with a primary spoke PW and a standby spoke PW linked to separate N-PEs. The standby PW is used by the U-PE in the event that the primary route fails. The new N-PE may use a MAC List TLV to send a MAC Flush message to all other N-PEs, telling them to remove any outdated MAC associations for that VPLS instance in order to expedite convergence.
You can also read CCNA Topics And What Is A CCNA Certification In Networking?