Access Networks and Core Networks

Core networks have relatively few network elements (routers, LSRs, switches), while access networks have many (CPEs, NTUs, DSLAMs, aggregators). This means that there is stronger pressure on access network CapEx, and that access network elements need to be as “touchless” as possible. The core runs higher data rates while the access runs lower data rates (including DSL, PON, wireless). Thus the core may guarantee Quality of Service (QoS) by resource overprovisioning, while the access needs true QoS mechanisms (such as token bucketing).

The core is richly connected, while access network topologies are typically tree or ring configurations. Thus a fault in the access network affects fewer people than a fault in the core, but there are fewer bypass options. The core can get away with fast re-route (FRR) while the access network requires OAM and planned Automatic Protection Switching (APS). The core network elements are in well-guarded networking installations while access network elements are often readily accessible to outsiders. Thus, the core can be considered a “walled garden” from a security point of view, since it features strong security to and from the outside world but loose security on the inside. While customer networks can also be considered walled gardens, it is impractical to protect the entire access network.

Ethernet and MPLS-TP

While both Ethernet and MPLS are commonly used to carry IP, there are many fundamental protocol differences between the two. Ethernet is defined from Layer 0 to Layer 2 (but may run over MPLS), while MPLS always requires a foreign server layer to transport it (which may be Ethernet). Ethernet frames are inherently self-describing, while MPLS packets do not contain a protocol ID. Every Ethernet frame contains a global non-aggregatable destination address, but MPLS packets have only locally-meaningful labels. Every Ethernet frame contains a unique source address, but MPLS packets contain no source identifier.

Both Ethernet and MPLS-TP can transport IP and other clients. Both Ethernet and MPLS-TP can be transmitted over SDH/SONET and OTNs (Optical Transport Networks). Both Ethernet and MPLS-TP define fault management and performance management OAM, as well as APS mechanisms. Ethernet does not define a routing protocol (neglecting TRILL and similar recent proposals) but defines a number of Layer 2 control protocols (L2CPs). On the other hand MPLS leverages the entire IP suite of protocols. Ethernet does not tolerate topology loops, while MPLS, having a Time To Live (TTL) field, can survive transient loops. Ethernet and MPLS both define 3-bit priority (DiffServ) marking. S-tagged Ethernet
also supports Drop Eligibility marking. Carrier-grade Ethernet supports bandwidth profiles (token bucketing). Ethernet defines timing (1588) and security (MACsec) protocols. For both protocols a single entity claims to hold the pen for the specification – the IEEE for Ethernet and the IETF for MPLS. However, in actuality, multiple competing Service Data Objects (SDOs) work on development of both.

Download full article PDF: 24001_Ethernet-MPLS-TP-in-Access-WhitePaper