Anda di halaman 1dari 10

T E C H N O L O G Y W H I T E PA P E R

Metro Ethernet Quality of Services

We describe steps required for realizing QoS for the metro Ethernet services, including classification of Customer-VLAN (C-VLAN) traffic, relevant traffic contract specifications per CVLAN at the User to Network Interface (UNI), appropriate buffer management and scheduling of the queues at the nodes, and admission control and traffic engineering geared towards multipoint services. General functional requirements such as provider edge classification, mapping of traffic management fields, scheduling, shaping, buffer management, congestion control, and performance monitoring and OAM are also addressed.

B. Raahemi, G. Chiruvolu, A. Ge, M. Ali

METRO ETHERNET QUALITY OF SERVICES


A framework for quality of services in metro Ethernet networks: the essential functional blocks and handling of packets in the core and at the provider and customer edges.

Transporting Ethernet using Synchronous Optical NETwork / Synchronous Digital Following the dominance of Ethernet in local area, Hierarchy (SONET/SDH) via Generic Framing campus and enterprise networks, efforts are underway in Procedure (GFP) encapsulation and Link Capacity the Institute of Electrical and Electronics Engineers Adjustment Scheme (LCAS) rate adaptation at the (IEEE), the Metro Ethernet Forum (MEF), and the edge of the transport network. International Telecommunications Union (ITU) to extend Using generalized MPLS to control Ethernet Ethernet into metro networks. Service providers are switches in the metro network. already starting to deploy scalable, cost-effective valueadded metro services. Although Ethernet technology now supports different traffic priorities (voice, video, data) in The first two are the focus of this article. enterprise networks, the major challenge for deploying this technology in metro networks and Wide Area MPLS metro networks Networks (WAN) is to guarantee the Quality of Services As illustrated in Figure 1, the metro network comprises (QoS). Other challenges include scalability (Ethernet flat Provider Edge (PE) routers, Label Switching Routers addressing), traffic engineering (load balancing), reliability, and Fig. 1 Network scenario 1: MPLS metro network Operations, Administration and Maintenance (OAM) features. Tunnel VC MAC MAC VLAN Eth Data CRC Label Label DA DA Tag Type A number of basic building blocks are required to guarantee Frame Format PE CE-B QoS in metro Ethernet networks, including provider edge LSP1 UNI-B classification, mapping of traffic LSR1 management fields, scheduling, shaping, buffer management, CE-A PE LSR3 congestion control, and Metro Ethernet performance monitoring and OAM. Network
UNI-A

Introduction

Network Scenarios
Several architectures can be used to carry Ethernet frames across metro networks: Using Multi Protocol Label Switching (MPLS) as the transport technology in the metro network. Extending the native Ethernet protocol, currently under consideration in IEEE 802.1 [1].

LSP2 LSR2

LSP3

PE UNI-C CE-C

CRC: Cyclic Redundancy Check DA: Destination Address SA: Source Address

2 | Alcatel Telecommunications Review - 4 th Quarter 2004

METRO ETHERNET QUALITY OF SERVICES

(LSR) and Label Switched Paths (LSP) between the two PE routers. The PEs are connected to Customer Equipment (CE) via the User to Network Interface (UNI). In this scenario, MPLS layer 2 encapsulation (also known as Martini encapsulation) facilitates the transportation of layer 2 frames across an MPLS service provider domain [2,3]. Two MPLS labels are inserted into the customer Ethernet frames based on destination Media Access Control (MAC) address/ port/802.1Q information at the ingress nodes. The first label at the top of the stack is the tunnel label, which is used to carry the frame across the provider network. The core LSRs only look at the top label to switch the labeled frame across the MPLS domain. The top tunnel label is typically removed by the penultimate hop, that is, the hop before the egress Label Edge Router (LER). The second label at the bottom of the stack is the Virtual Circuit (VC) label, which is used by the egress LER to determine how to process the frame and where to deliver it on the destination network. The egress LER infers from the VC label how to process the frame and then forwards it to the appropriate outgoing port. As the VC label is not visible until the frame reaches the egress LER because of the MPLS tunneling hierarchy, both the tunnel and VC labels are required in the case of MPLS encapsulation.

Fig. 2

Network scenario 2: Provider bridged network


ID CFI VLAN ID (3 bits) (1 bit) (12 bits)

MAC DA

MAC SA

Eth. Type 0x8100

1Q pVLAN

1Q Eth VLAN Type Tag

Orig. Eth. Type

Data PE

CRC CE-B UNI-B

Frame Format

CES CE-A PE CES Metro Ethernet Network CES Spanning Tree PE UNI-C CE-C

UNI-A

CES: Core Ethernet Switch / Bridge CFI: Canonical Format Indicator ID: IDentifier P-VLAN: Provider VLAN

Fig. 3

E-Line service type using point-to-point EVC

CE

PE UNI

Metro Ethernet Network Point-to-Point EVC

PE

CE

UNI

Provider bridged networks This scenario, which is an extension of the native Ethernet protocol into metro networks, is currently being considered in IEEE 802.1 [1]. As shown in Figure 2, the metro network comprises Ethernet switches/bridges. A spanning tree protocol is used to establish a path between the PEs for each customer Virtual Local Area Network (VLAN). An encapsulation scheme, such as Q-in-Q (VLAN stacking) or Mac-in-Mac [4], is used to transport customer traffic across the metro domain.

[5]. The customer equipment is connected to the metro Ethernet network via its Ethernet port. This UNI acts as the physical demarcation point between the service provider and subscriber domains. Inside metro networks, the connectivity between UNIs is provided by an Ethernet Virtual Connection (EVC). A transport layer, such as SDH, Ethernet, MPLS or Asynchronous Transfer Mode (ATM), provides this virtual connection. From the customers viewpoint, the metro network looks like an Ethernet wire. E-Line and E-LAN services Two types of services are defined based on the network connectivity: Ethernet Line (E-Line) and Ethernet LAN (E-LAN) services [6]. The E-Line

Metro Ethernet Services


The MEF has defined a number of new metro Ethernet services based on the network architecture of Figure 3

Alcatel Telecommunications Review - 4 th Quarter 2004 | 3

METRO ETHERNET QUALITY OF SERVICES

service, which provides a point-to-point EVC between two UNIs, can be used to create a broad range of pointto-point services, such as Ethernet Private Line (EPL) services and Ethernet Virtual Private Line (EVPL) services. There are more sophisticated forms of E-Line service [6]. Service multiplexing of more than one EVC may occur at none, one or both of the UNIs. Figure 4 shows an example of the EVPL service in which one physical port at customer A (UNI A) supports connectivity for two Ethernet services. Some service frames may be sent to EVC1 (Customer B) while other service frames may be sent to EVC2 (Customer C).

Essential Building Blocks for Metro Ethernet QoS

Fig. 4

Quality of service is an important aspect of Ethernet services. Ethernet traffic is carried over an EVC, which might use a tunneling technology, such as MPLS or Qin-Q, with the tunnels being set up a priori or ondemand. Therefore, the QoS requirements should be met separately for the tunnels and the traffic that is being carried through these tunnels. Depending on the network design, the tunnel QoS requirements can differ between E-Line and E-LAN services. It is assumed that the QoS requirements for the tunnels and specification are met separately from the customer traffic. Based on [7], this article elaborates on the minimum set of functions required EVPL service using E-Line service type to support QoS in metro Ethernet networks. Several steps need to be taken to PE CE-B realize QoS for Ethernet services, including: classification of UNI-B EVC1 Customer VLAN (C-VLAN) traffic; CE-A PE specification of suitable traffic Metro Ethernet contracts per C-VLAN at the UNI; Network buffer management and scheduling UNI-A PE of the queues at the nodes; and CE-C EVC2 admission control and traffic engineering geared towards UNI-C multipoint services.

Provider edge classification Several C-VLAN configurations are possible, depending on the Fig. 5 LAN extension using an E-LAN service type and logical grouping of endstations. Grouping of users can be based on MAC addresses, IP Edge Bridge CE-B subnets and other criteria to Metro Ethernet determine the C-VLAN Network UNI-B membership. In metro Ethernets, CE-A Edge Bridge the CE assigns the C-VLAN EVC1 IDentifiers (VID). The PE uses UNI-A these VIDs to determine to which PE CE-C EVC the frame belongs. In turn, the EVC is determined by the stacked Q-tag VID in the case of UNI-C Edge provider bridges or by the tunnel Bridge label in the case of Virtual Private LAN Service (VPLS) schemes which are based on Ethernet over MPLS (EoMPLS). In the EoMPLS case, the Class of Service (CoS) and any other To connect three or more sites, the subscriber could attributes are inferred either from the label itself, or use the E-LAN service for which the carrier network explicitly from the EXPerimental (EXP) bits within the performs switching/bridging functions. Figure 5 MPLS label. The incoming frames at the UNI are illustrates LAN extension using an E-LAN service. The classified at the provider edge and are thus mapped/ edge nodes perform a bridging/switching function, while a encapsulated appropriately and transported across the spanning tree protocol is used to prevent loops in the appropriate EVCs. provider network.

4 | Alcatel Telecommunications Review - 4 th Quarter 2004

METRO ETHERNET QUALITY OF SERVICES

Ingress mapping of traffic management fields In the MPLS-based core network scenario, the edge node maps the customer traffic (e.g. learned MAC addresses / VIDs, ports) onto an LSP that has been preestablished by the MPLS forwarder; this facilitates the transport of the (Martini) encapsulated frames across the metro domain. Label Distribution Protocol (LDP) signaling is required to distribute the labels, while LSRs are needed to determine the routes in order to establish the required LSPs. In the provider bridge network scenario, the stacked .1q tag has the same function as the MPLS label (i.e. enabling the segregation of more than 4096 customers). Each scheme has its own advantages and disadvantages [4]. The ingress provider edge may consider the user priority (.1p field) [1] of the VLAN tag header in order to determine the QoS field of the outer label of the LSP (e.g. the EXP fields of the MPLS label stack). Note that the customer QoS markings may or may not be used to determine the outer QoS bits of the provider tag. Several such mappings are possible; the mappings should provide appropriate qualities of service for the various 802.1d classes of service [2]. In the case of provider bridge networks, the .1p information should be appropriately carried in the .1q tag. These methods provide appropriate CoS mappings onto the EVCs; the EVC classification identifier is used to map the EXP field. The IEEE 802.1ad group is currently defining the relevant drop precedence and mapping for backward compatibility. Encoding drop precedence into p-bits (Ethernet VLAN priority bits) could ensure a more robust QoS and flexible bandwidth profile in the Service Level Agreement (SLA). Alcatel has presented a proposal to the MEF for a default mapping with two options (using 3 or 4 queues). Table 1 shows such a default mapping using three queues and three classes of services. A default mapping maximizes the plug-and-play aspect of Ethernet and its backward compatibility with IEEE 802.1d switches.

Egress mapping of traffic management fields In this article it is assumed that the traffic management functions are uniquely mapped and understood among the customer sites. With this assumption, generally no special egress mapping function is needed as the core configured tunnels terminate at the provider edge and, depending on what device is used at the customer premises [Customer Equipment / Customer Located Equipment (CLE) / Multi-Tenant Unit (MTU), etc], in provider bridge networks user traffic to the customer is carried either directly or via the inner tunnel (e.g. Martini or MPLS label) that already has the traffic management fields as marked by the source. It should be noted that if the CE-VID is not preserved across the metro domain, the egress node restores the original customer VIDs as negotiated during the setup phase. Similar functions are required with respect to CoS. Traffic contracts Traffic contracts for services need to be specified perUNI, per-EVC and per-CoS. A combination of profiles can be specified. The traffic contract for an EVC of an E-Line service is specified as ingress/egress bandwidth profiles. The bandwidth profile can be specified using the MEF specification [8] or the Internet Engineering Task Force (IETF) specification [9,10]. Frames are marked with appropriate colors indicating the priorities with the help of Token Bucket [8]. Nodes in the network perform different congestion control schemes (e.g. Weighted Random Early Detection; WRED) for various traffic colors. Re-mark assumes that the service is color-aware for the user traffic and recognizes user traffic color marking, whereas in the case of mark it is color-blind. Traffic shaping depends on the capability of the equipment and need not be compulsory. It can be performed at the network ingress, where it is used for soft policing. Alternatively, shaping can be performed at the network egress towards the customer; it can be used when multiplexed traffic from various EVCs (e.g. E-LAN service) exceeds the customers egress traffic profile. Generally, two service models can be deployed for a multipoint service: pipe model and hose model. In the pipe model, the service contract and bandwidth profile specification relate to each pair of communicating sites; point-to-point EVCs are set up for this purpose. In the hose model, the bandwidth profile is specified as a whole between the customer site and the network. The traffic distribution between each pair of communicating sites is not specified. This model eases the bandwidth profile specification, but complicates network design and resource allocation within the network. Therefore, for an E-LAN service, the traffic contract specification depends on the service model. In both cases, the bandwidth profile (per-hose or per-pipe) and the treatment of bandwidth profile violations must be included in the traffic contracts.

Tab. 1

Possible default mapping of drop precedence and class of services into p-bits using three queues
CoS + drop precedence identifier value 7 6 5 (Green color) 4 (Yellow color) 3 0 2 1 p-bits 111 110 101 100 011 000 010 001 802.1d recommende mapping (3 queues) Voice Control Load Best Effort 7 6 5 4 3 0 2 1

Forwarding class name premium gold

standard

Alcatel Telecommunications Review - 4 th Quarter 2004 | 5

METRO ETHERNET QUALITY OF SERVICES

Ingress traffic control As discussed earlier, a service model associated with a given service (E-Line or E-LAN) binds a traffic contract to the network. The network operator may use traffic policers to check for violations in service contracts. Depending on the traffic contract parameters, a single rate Three Color Marker (srTCM) [9] or a two rate Three Color Marker (trTCM) [10] can be employed for the IETF specification. In the case of the MEF specification, a simple token bucket based rate/burst profiler can identify conformant and nonconformant traffic [8]. The traffic policer checks for bandwidth profile violations and takes appropriate action in accordance with the service contract. Possible actions are: drop traffic that violates the traffic contract, mark traffic with a color of lower precedence if the network uses different colors to identify different traffic priority precedence in a non-color-aware mode, or remark the traffic with a color of lower precedence if the user traffic is already colored and the traffic contract specifies that the user coloring scheme should be recognized. If the network equipment can shape the traffic into the network, the traffic contract can also specify the shaping option. Scheduling Traffic belonging to multiple classes needs to be appropriately scheduled onto the network links in order to meet the required service guarantees. The network equipment must be capable of supporting various traffic scheduling schemes, such as Strict Priority, Weighted Round Robin (WRR) and Weighted Fair Queuing (WFQ), to accommodate the service guarantees. Generally, a service that requires strict delay guarantees is scheduled with the highest priority. If the scheduler only implements a WFQ or WRR scheme, then the service is given the largest possible weight. All other services are given appropriate weights in line with the traffic contract. For example, the weights can be set proportional to the sum of the equivalent bandwidths of all the services being transported by that particular traffic class, where the equivalent bandwidth is computed based on the ingress bandwidth profiles. Alternatively, the weights among the traffic classes are configured a priori in the network and admission control ensures that the admitted traffic is within the allocated bandwidth limits for a given traffic class. Irrespective of the admission control scheme, nodes should include a weighted scheduler that differentiates queues and schedules the traffic based on the priority/traffic class. In some scenarios, more flexible scheduling is required at the service ingress and/or service egress. The Alcatel 7750 Service Router and Alcatel 7670 Switch/Router support hierarchical scheduling whereby a parent scheduler can be created for a bundle of service queues (say gold, silver and bronze services) that limits the overall rate of all the queues. Customers are able to send in any combination of gold, silver and bronze traffic conforming to their specified

Peak Information Rate (PIR) values that does not exceed the maximum PIR for the three services. Buffer management and congestion control Buffer management schemes address the issues of transient congestion and priority-based preferential packet dropping in order to achieve relative QoS. These functions differ between the core and edge nodes. At the edge nodes, IEEE 802.3x Pause [11] can be deployed to alleviate congestion. Alternatively, preferential packet drop can be deployed utilizing p-bits or EXP bits in the tags. In the core, packet-dropping mechanisms, such as WRED, can alleviate/avoid network congestion. At the edge, the IEEE 802.3x Pause control can be useful as a congestion avoidance mechanism in VLAN services or when the ingress PE cannot process the frames arriving at the UNI at wire speed. In 802.3x, the input buffer is constantly monitored at an Ethernet interface; when the packets/frames in the buffer exceed a certain threshold value (i.e. congestion is detected), the congested node sends a pause message to the upstream node. Upon receiving the pause message, the upstream node stops transmitting packets for the duration indicated in the pause message. An extension of 802.3x, namely DiffPause [12], is proposed to provide fair, scalable and color-aware congestion control in metro Ethernet. The scheme takes advantage of a per-link based backpressure mechanism without any bandwidth reservation; it is independent of the number of ongoing individual sessions Egress traffic shaping Depending upon the routing and traffic patterns, when multiple EVCs terminate at a UNI, the total egress traffic towards the customer may exceed the egress traffic profile. In this case, the edge node may have to perform per-UNI congestion control. Traffic can be limited either by policing/dropping or by shaping using buffers at an egress node. Shaping may require large buffers (and buffer management schemes) in contrast to policing/dropping. However, shaping can reduce both frame loss and traffic burstiness. Profiling can be per class of service, per CVLAN, or per aggregated bundle (group of C-VLANs). Network traffic engineering Multipoint traffic engineering is a key issue for efficient resource allocation and load balancing in the metro Ethernet [13]. Each C-VLAN can access the metro domain via multiple PE nodes through pre-established multipoint EVCs. In addition, multiple spanning trees can provide alternate routes for C-VLANs across the provider network. Furthermore, C-VLAN grouping makes it possible to support a large number of C-VLANs in the provider network. Multipoint traffic engineering schemes can benefit from such grouping schemes in terms of reduced computation. SLAs with the customers are translated into resource allocation requirements. Appropriate signaling mechanisms should be devised to meet the QoS for multipoint EVC

6 | Alcatel Telecommunications Review - 4 th Quarter 2004

METRO ETHERNET QUALITY OF SERVICES

services. For example, the IEEE General Address Registry Protocol (GARP) needs to be extended to support multipoint EVC services in the case of provider bridges; either Resource Reservation Protocol (RSVP) or LDP signaling can be used for the MPLS domain. It should be noted that VPLS schemes greatly benefit from the advanced features that are traditionally associated with MPLS, such as bandwidth guarantees and traffic engineering through appropriate signaling mechanisms, such as RSVP Traffic Engineering [14].

Fig. 6

Desired functions at different network elements to support QoS


at Egress PE at ingress PE - Scheduling/Buffer Mngmnt - Admission Control - Frame Marking - Traffic Mapping at Core P - Scheduling/Buffer Management - Traffic Shaping - Scheduling/Buffer Mngmnt - Traffic Mapping

at CE - Traffic Shaping - Admission Control - Frame Marking

PE UNI-B

CE-B

Core CE-A PE UNI-A Metro Ethernet Network Core

Admission control Admission control is another critical component for realizing the desired QoS for various classes/ services. It estimates the available network resources (such as bandwidth) and decides on-demand whether a customers request for bandwidth can be met. In MPLSbased core networks, nodes reserve the necessary resources through signaling. Provisioning of new services will be based on the available network resources; attempts to set up a new EVC setup may be rejected if there are insufficient resources. In the case of provider bridge networks, a network management system with bandwidth broker type functionality may be needed to admit / reject the provisioning of new services. As described earlier, customer traffic can be aggregated into service provider tunnels, in which case a hierarchical admission control function is needed: a first level checks the bandwidth requirements for service provider tunnels, while a second level is provided at service multiplexing points. Figure 6 illustrates the functions needed at each network element to support QoS in a metro network. Performance monitoring and SLS verification The performance parameters describe the service guarantees the network offers to the customer for the packet stream described by the flow description. There are four major performance parameters: delay, jitter, packet loss and throughput. Other parameters of interest are availability, service schedule, monitoring and reporting. The availability of the offered network service needs to be related to the network protection and restoration options. The service schedule indicates the start and end times. Finally, the monitoring and reporting section of the Service

Core PE UNI-C CE-C

Level Specification (SLS) specifies when and how the QoS performance needs to be monitored and reported. OAM OAM is another key element in ensuring that the provider network meets the contracted SLA. OAM tools are currently being developed by several standardization bodies, including the IETF [15], ITU [16], IEEE [17] and MEF [18]. These will provide powerful diagnostic functions to support QoS in metro Ethernet. The Ethernet OAM protocols should be able to determine the connectivity, measuring the round-trip delay and jitter, and providing alarm notification and performance measurements. These OAM functions facilitate the collection of essential information on routing, delays and QoS [19].

Conclusion
Service providers need to consider a crucial revenue generating element in metro networks, that is, the ability to guarantee QoS in conformance with the SLAs for various services, such as E-Line and E-LAN services. This article presents a comprehensive framework for supporting QoS in metro Ethernet networks, including an overview of the main functional blocks (e.g. traffic shaping and policing, admission control, scheduling and buffer management).

Alcatel Telecommunications Review - 4 th Quarter 2004 | 7

METRO ETHERNET QUALITY OF SERVICES

It also describes how packets are handled at different network elements (CE, PE, and core). Table 2 summarizes the desired functions. Tab. 2
Attribute Traffic Shaping Scheduling/ Buffer Management Admission Control Frame Marking Traffic Mapping

Summary of the required functions at various network elements


CE Yes No PE (ingress/egress) Yes Yes Core No Yes

Yes Yes No

Yes Yes Yes

No No No

Alcatels family of Service Routers (Alcatel 7750 and 7450)[20,21], Optical Multi-Service Nodes (OMSN), Multi Service Provisioning Platforms (MSPP) (Alcatel 1660 SM) [22], and multi-service system for customer premises (Alcatel 1642 Edge Multiplexer) [23], supports a wide variety of metro services and provides a comprehensive range of OAM and provisioning features. Service providers require these features to implement QoS and to deliver guaranteed SLAs in their metro Ethernet networks, as well as to provide solutions to the challenging problems that carriers and service providers face today.

References
[1] IEEE 802.1D/Q/p Working Group: Media access control (MAC) bridges, IEEE 802.1D, 1998. [2] http://www.ietf.org/internet-drafts/draft-ietf-pwe3ethernet-encap-04.txt. [3] M. Lasserre, V. Kompella: Virtual Private LAN Services over MPLS, IETF Work in Progress, draft-ietf-l2vpn-vplsldp-03.txt, April 2004. [4] G. Chiruvolu, B. Krogfoss, A. Ge: Encapsulation Schemes to Extend Ethernet to Metropolitan Area Networks: A Comprehensive Analysis of Popular and Evolving Encapsulation Schemes for Metro-Ethernet, White Paper, Alcatel 2004. [5] Ethernet Services Model, Phase I, Metro Ethernet Forum, November 2003.

[6] Ethernet Services Definitions Phase I, Draft v5.5, Metro Ethernet Forum, March 2004. [7] Functional Description of Quality of Services in Metro Ethernet networks, Draft 1, Alcatel Internal Document (produced by Alcatel Metro Ethernet QoS Task Force), April 2004. [8] Traffic Management Specification-Phase I, draft. V7.0, Metro Ethernet Forum, March 2004. [9] A Single Rate Three Color Marker, RFC 2697, IETF. [10] A Two Rate Three Color Marker, RFC 2698, IETF. [11] IEEE Standards 802.3x on congestion control 1997. [12] A. Ge, G. Chiruvolu: DiffServ-Compatible Fair Congestion Control through Extended Pause (DiffPause) for Metro-Ethernet, IEEE ICC, 2004. [13] H. Saito, Y. Miyao, M. Yoshida: Traffic engineering using multiple multipoint-to-point LSPs INFOCOM 2000, Proceedings. IEEE, 26-30 March 2000, volume 2, pp 894-901. [14] D. Awduche et al: RSVP-TE: Extensions to RSVP for LSP Tunnels, RFC 3209, http://www.ietf.org/rfc/rfc3209.txt. [15] M. Aissoui et al: OAM Procedures for VPWS Interworking, IETF Draft, Work In Progress, 2004. [16] Study Group 13, ITU, Work in Progress. [17] N. Finn: Bridges and End-to-End OAM, IEEE 802.1 work in progress, March 2003 http://www.ieee802.org/1/files/public/docs2003/finne2e-oam-bridges-1.pdf. [18] Metro Ethernet Forum, http://www.metroethernetforum.org/. [19] A. Berthillier, A. Lardies: Ethernet OAM status review, Alcatel position paper (internal document), 2004. [20] Alcatel 7750 Service Router, http://www.alcatel.com/products/productsummary.jhtml? repositoryID=/x/opgproduct/a7750sr.jhtml. [21] Alcatel 7450 Ethernet Service Switch, http://www.alcatel.com/products/productsummary.jhtml? repositoryID=/x/opgproduct/a7450ess.jhtml. [22] Alcatel 1660 SM STM-16/64 Optical Multi-Service Node for Metro Applications, http://www.alcatel.com/products/productsummary.jhtml? repositoryID=/x/opgproduct/Alcatel_1660_SM.jhtml. [23] Alcatel 1642 Edge Multiplexer, http://www.alcatel.com/products/productsummary.jhtml? repositoryID=/x/opgproduct/Alcatel_1642_em.jhtml.

Bijan Raahemi is a Research Scientist at the Alcatel Research & Innovation Center in Ottawa, Canada. He has contributed to the Ethernet Services definition at the Metro Ethernet Forum. (Bijan.Raahemi@alcatel.com) Girish Chiruvolu is a Research Scientist in the Packet Transport and Interworking Group at the Alcatel Research & Innovation Center in Plano, Texas, USA. (Girish.Chiruvolu@alcatel.com)

An Ge is a Research Scientist in the Packet Transport and Interworking Group at the Alcatel Research & Innovation Center in Plano, Texas, USA. (Andrew.Ge@alcatel.com) Maher Ali is a Research Scientist in the Packet Transport and Interworking Group at the Alcatel Research & Innovation Center in Plano, Texas, USA. (Maher.Ali@alcatel.com)

8 | Alcatel Telecommunications Review - 4 th Quarter 2004

METRO ETHERNET QUALITY OF SERVICES

Abbreviations
ATM CAC CBS CE CES CF CIR CLE CoS CRC C-VLAN DA EBS EIR E-LAN E-LINE EoMPLS EPL EVC EVPL EXP GARP GFP IEEE IETF ITU LAN LCAS LDP LER LSP LSR MAC MEF MPLS MSPP MTU OAM OMSN P PE PIR QoS RSVP SA SDH SLA SLS SONET srTCM trTCM UNI VC VID VLAN Asynchronous Transfer Mode Connection Admission Control Committed Burst Size Customer Equipment Core Ethernet Switch / Bridge Coupling Flag Committed Information Rate Customer Located Equipment Class of Service Cyclic Redundancy Check Customer VLAN Destination Address Excess Burst Size Excess Information Rate Ethernet LAN services Ethernet Line services Ethernet over MPLS Ethernet Private Line Ethernet Virtual Connection Ethernet Virtual Private Line EXPerimental General Address Registry Protocol Generic Framing Procedure Institute of Electrical and Electronics Engineers Internet Engineering Task Force International Telecommunications Union Local Area Network Link Capacity Adjustment Scheme Label Distribution Protocol Label Edge Router Label Switched Paths Label Switch Routers Media Access Control Metro Ethernet Forum Multi Protocol Label Switching Multi Service Provisioning Platform Maximum Transmission Unit Operations, Administration and Maintenance Optical Multi-Service Node Core Provider Edge Peak Information Rate Quality of Services Resource Reservation Protocol Source Address Synchronous Digital Hierarchy Service Level Agreement Service Level Specification Synchronous Optical NETwork single rate Three Color Marker two rate Three Color Marker User to Network Interface Virtual Circuit VLAN IDentification Virtual Local Area Network

VPLS WAN WFQ WRED WRR

Virtual Private LAN Service Wide Area Networks Weight Fair Queuing Weighted Random Early Discard Weight Round Robin

Alcatel Telecommunications Review - 4 th Quarter 2004 | 9

Alcatel and the Alcatel logo are registered trademarks of Alcatel. All other trademarks are the property of their respective owners. Alcatel assumes no responsibility for the accuracy of the information presented, which is subject to change without notice. 11 2004 Alcatel. All rights reserved. 3GQ 00009 0005 TQZZA Ed.01

Anda mungkin juga menyukai