|
Archive 28 Feb 2007 #57Ethernet's New Frontier: "Self-Insured" Ethernet By Vladimir Kaminsky, Sr. Consultant This is Part I from a two-part paper prepared by Flanagan Consulting on an emerging Ethernet protocol whose goal is to compete with MPLS and Ethernet- over-SONET transport techniques for Carrier-Grade transmission. Though there are many questions still awaiting resolution, it looks like Provider Backbone Transport (PBT) may not require additional technologies to keep Ethernet transmission performance parameters (such as resilience, OA&M, and others) on the Carrier level. Even without final approval of standards, several vendors (Nortel, Extreme, and others) produce equipment based on this protocol. British Telecoms is one of the first major service providers to agree to test it in a wide deployment. Part I: Out of Touch with Reality Ethernet, a simple and now ubiquitous data protocol, originally was designed for the LAN environment. As a result, Enet might have suffered in its popularity if other transport technologies had not come to its rescue and supplied important features that Ethernet lacks. For example, in building Carrier-Grade networks (Enterprise, MAN, Core) Enet on its own has issues with: *OA&M features: today's users requirements, though well supported, do not correspond to what a carrier needs; *Network protection: mechanisms and speed of reaction to failures are not adequate; *Size of the network is limited; *There are no End-to-End QoS guarantees; *Low scalability and network resource utilization; *Problems with latency and jitter. These factors, together with many known Ethernet advantages urged the industry to develop methods to mitigate Ethernet problems, and to look for advanced transport methods. So far, there is no unique solution, and various techniques have been proposed to make Ethernet more efficient. Encapsulation of Ethernet frames into another signal, which is called "a carrier" in general, is a common method for many of these techniques. Next Generation SONET (NGS) The schemes to adopt SONET for Ethernet transport are currently developed and implemented. Some of them are applicable not only for Ethernet, and a loosely defined term "Next Generation SONET" is used to describe SONET with enhancements to transport data. Practical applications found: * Virtual concatenation (VCAT); * Link capacity adjustment scheme (LCAS). VCAT and LCAS define the methods for transport (Multiplexing Layer); *Generalized Multiprotocol Label Switching (GMPLS); *Generic framing procedure (GFP). The GFP-framed and GFP-transparent version is standardized, and the GFP-transparent version is still in a process of standardization; *Link access procedure for SDH (LAPS). GFP and LAPS are layer 1 adaptation protocols for transport. The standard bodies continue working on the development of these techniques. The ITU-T Standards Committee Study Group 15 issued a number of recommendations: *ITU G.7041/Y1303, January 2002: Generic Framing Procedure; *ITU G.7042/Y1305, November 2001: LCAS for Virtually Concatenated Signals; *ITU Recommendation X.85/Y.1321, March 2001: IP over SDH Using LAPS; *ITU Recommendation X.86, February 2001: Ethernet over LAPS; *Internet Draft "Framework for GMPLS-based Control of SDH/SONET Networks. Some of these techniques complement each other and have found applications working together. Ethernet, when it is transported with the help of NGS, is called "Ethernet-over-SONET" (EoS). Generalized Multiprotocol Label Switching GMPLS extends the Multiprotocol Label Switching (MPLS) concept to TDM and other transport environments. Among various applications, GMPLS is used to dynamically modify the bandwidth of a TDM (SONET) Label Switched Path (LSP). GMPLS defines how to use multiple LSPs routed diversely, to build a single SONET VCAT circuit. In this scheme, each component LSP can be protected and restored individually, offering great flexibility in planning for resiliency and high availability. The dynamic bandwidth allocation goal is to align bandwidth usage with "instantaneous" Ethernet rate. According to the demand, the size of a concatenated SONET connection can be modified dynamically, without having to reestablish a new connection or interrupt the service. These and several other methods to improve resiliency and operational characteristics of native Ethernet are not obtained for free. Improvements add additional layers of complexity to data networks, but have been necessary for Carrier-Grade transmission characteristics such as failure recovery in less than 50 ms. Despite the increased complexity, Carrier-Grade Ethernet is now one of the most popular data formats from LANs to network core. Multiple providers offer services to support such networks, and they have proved to be highly reliable and attractive in the wide range of data rates (from T1 to OC-48). At the same time, as mentioned, they require a combination of several transmission techniques that increases the cost and complexity; that is why researchers tried to enhance Ethernet itself to develop new protocols to mitigate native Ethernet weakness. In Part II, we will introduce one of such protocols, Provider Backbone Transport, that, in our opinion has a promising future. "Flanagan
Consulting" and "ViewsLetter" are Service Marks of W. A. Flanagan, Inc.
|
![]() W. A. Flanagan, Inc. 45472 Holiday Drive, #3 Sterling, VA 20166 Ph: +1.703.242.8381 Fx: +1.703.242.8391 |
In Converged Networking, We have the Experience Editor@ViewsLetter.com www.flanagan-consulting.com [directions] |