Bandwidth provisioning between an MTU and a PE is extremely difficult with a full-mesh VPLS design.
Since each VPLS instance can require multiple tunnel LSPs, the bandwidth requirements for each tunnel LSP must be separately accepted and individually enforced by every PE a tunnel LSP traverses. Because the provider requirement is to manage the provisioned bandwidth for the VPLS and not each tunnel LSP, the MTU has the added responsibility of rate limiting the aggregate egress traffic across multiple tunnel LSPs on the uplink. Due to packet replication issues described previously, this is not practical.
Hierarchical VPLS designs simplify bandwidth provisioning and management. Because tunnel LSPs from the MTU are terminated at the PE, tunnel LSP resources are easily shared and managed between customers. Thus, traffic for multiple VPLS instances can be transported across a single tunnel LSP. In some cases only a single best-effort tunnel LSP is required between the MTU and the PE. Traffic for each customer is carried over a different pseudowire on the same tunnel LSP. This allows the tunnel LSP to be signaled once, with the desired bandwidth and priority parameters sufficient for providing best-effort service for customers connected to the spoke peer. If a customer upgrades their service or a new customer is connected that requires guaranteed bandwidth, a second tunnel LSP could be signaled with the SLA bandwidth parameters. Once established, the second tunnel LSP can carry traffic for a single customer as a premium service.