Carrier Ethernet and Provider Backbone Bridging
The basic separation required for public service networks can be achieved using VLANs, however the conventional IEEE 802.1q approach still has many shortcomings when applied to carrier networks. This type of VLAN separation is called a “Q-tag”, because of the standard which defines it. In fact up to three levels of hierarchy are required to implement an end-to-end Ethernet service effectively:
- Customer VLANS for separate departments within the customer organisation must be transported
- VLANs must be used to separate service instances in the access network, typically one per customer
- VLANs must be used within the service provider core network to maintain security and separation
IEEE 802.1ad is also known as “Q-in-Q”, stacked VLANs or Provider Bridges. It adds a second 802.1q VLAN ID (VID) so that the original customer VID can be carried transparently across the WAN, and the Service Provider VID can be used to separate customers. The original 12 bit VLAN address space of 802.1q is maintained, so that no more than 4096 service instances can be supported. This limits the usefulness of the technology.
Provider Backbone Bridges (PBB), also known as “Mac-in-MAC” are standardized in IEEE 802.1ah. An additional service tag with a 24-bit address space is included and used to separate customers, leading to a more scalable solution. The use of a separate Medium Access Control (MAC) header for both service provider and customer also means there is a clearer and stronger demarcation between the customer and service provider networks. This can simplify the operation of the service provider switches, which do not need to learn or interpret customer addressing. In this respect, 802.1ah is closer to the MPLS approach of label stacking.
Resilience for Ethernet networks in the LAN has traditionally been provided using Spanning Tree Protocol (STP). However STP is unsuitable for MAN and WAN use, because of its relatively slow restoration time, and its inefficiency in the use of ports. Various refinements have been developed and deployed in pure Metro Ethernet solutions, including Rapid Spanning Tree Protocol (RSTP) and Multiple Spanning Tree Protocol (MSTP). Where Ethernet is carried natively in the access or Metro network, these mechanisms must be relied upon and may be sufficient. However in many cases, Ethernet is carried in another transport layer, which typically provides additional resilience and restoration options. Popular choices for Ethernet transport include:
- Ethernet over SDH/SONET – this typically uses Generic Framing Procedure (GFP) to provide fine-grained and adjustable capacity based upon Virtual Concatenation (VC), with capacity matched to the bandwidth requirements of individual Ethernet service instances.
- Ethernet over MPLS – this typically makes use of Pseudo Wires to provide point-to-point capacity across the MPLS network for each service instance, often supported by label stacking and MPLS traffic engineering in the core IP/MPLS network.
Bookmark this article
Share this article using the following sites: