Provider Backbone Bridging Networks A Highly Scalable VLAN (Multicast) Architecture

Transcription

1 Provider Backbone Bridging Networks A Highly Scalable VLAN (Multicast) Architecture Paul Bottorff, Mark Holness, Norival Figueira, Michael Chen, Dinesh Mohan, Glenn Parsons Version 2.0 Page 1

2 A Provider Bridge Scaling Solution Provider Backbone Bridging 802.1ad Interfaces Provider Backbone Bridge Network Provider Bridge Network N Provider Bridge Network Page 2

3 Ethernet Service Types MEF Ethernet Virtual Connections (EVCs) E-LINE Router Mesh Pt-Pt, Like Duplex Ethernet Any-to-any E-TREE Hub & Spoke Pt-MPt, Like EPON Ethernet, Root-to-Leaf and Leaf-to-Root E-LAN Multi-Site MPt, Like VLAN, Any-to-any Page 3

4 E-LINE Dominates Today E-LINE is a natural leased line replacement for subscribers Ethernet leased lines offer high bandwidth Lines provide bandwidth on demand Interfaces are compatible with off the shelf Ethernet switches/routers Best for router mesh E-LINE provides natural migration for carriers Consistent with current operations model Allows carrier equipment reductions Bill models can follow well understood FR services Current QoS models allow both traffic control and service monitoring of E-LINE service offerings Service OAM models for E-LINE are relatively straightforward Each E-LINE service instance requires 1 S-VLAN Page 4

5 E-TREE Ideal For ISP Connect E-TREE Future Service With Great Promise Useful as a multiplexed connection to an application service provider like an ISP Service is unlike traditional Ethernet since leaf nodes can not talk with each other E-TREE has deployment issues No clear billing model For instance if one leaf is disconnected is the circuit down? What is the distance of the tree? OAM management not fully understood QoS model non-existant, SLAs can only provide Best Effort Page 5

6 E-TREE S-VLAN Mapping E-TREE Hub & Spoke Hub Port Spoke Ports Pt-MPt, Like EPON Ethernet, Root-to-Leaf and Leaf-to-Root Each E-TREE service instance requires 2 S-VLANs Both S-VLANs comprising an E-TREE S-VLANs are unidirectional The S-VLANs of and E-TREE service instance are typically on the multiplexed on the same port Page 6

7 Some Carriers Will Use E-LINE in Hub and Spoke Arrangement E-LINE Hub & Spoke Hub Port Spoke Ports Hub port would usually be multipexed to allow the multiple Pt-Pt attachments. Each E-LINE is a seperate managed S-VLAN This arrangement allows use of E-LINE management, billing, and QoS Many more S-VLANs are required Pt-Pt Root-to-Leaf and Leaf-to-Root Page 7

8 E-LAN Many Future Applications E-LAN is deployed for broad connectivity in select network Interconnect of multiple corporate sites Multi-player gaming Ubiquitous any-to-any connectivity E-LAN has many future applications E-LAN has deployment issues Deployments are very spotty Unclear billing model How is availability defined? No definitions for QoS or performance measurement What is the distance of a E-LAN Unclear management models Unlike existing carrier service offerings Each E-LAN service instance is a single S-VLAN Page 8

9 Prototypical Major Metro Area Business Subscriber Population 100K-2M San Jose Yellow Pages ~100K businesses The SF Bay Area lists ~1M businesses Large Business Sites 500-5,000 Residential Subscriber Population 1M-20M Leased Line Density 10K-200K Roughly 1/10 Yellow Page Listings Application Service Provider Sites Large APSPs sites may service residental Page 9

10 Major MSA Networks Typical SP Access Business CLE Small Office Medium Office Large Office Network Scale >10,ooo Remotes >10,ooo CLEs >500 COs COs COs Metro Scale >4,ooo Remotes >1,ooo CLEs >50 COs >20 COs >4 COs Typical Metropolitan Serving Area MSA MSA example shown ASIA/PAC more CO/MSA Europe less CO/MSA Page 10

11 Support 1,000,000 Service Instances Must be able to support E-LINE service for leased line replacement for entire MSA This is the way Ethernet is entering the markets The objective is 200K E-LINE instances Must support E-LINE for APSP to Subscribers Not all service providers will allow E-TREE because of deployment problems The objective of an additional 200K E-LINE is adequate for transition until E- TREE Requirements for around 10K E-TREE instances Requires 20K S-VLANs Must support E-LAN for APSP and B-B Advanced peer applications Number of service instances speculative, however could be large Totals 200K E-LINE S-VLANs for leased line replacement 200K E-LINE S-VLANs for APSP 20K E-TREE S-VLANs? E-LAN Service Instances Designing Into A Corner Will Not Instill Confidence In Future Set Objectives to at least 1,000,000 service instances E-LINE, E-TREE, E-LAN E-LAN service will eventually become important for coupling small groups Allow E-TREE and E-LAN service scaling to at least 100,000 for future growth Page 11

12 Provider Backbone Bridge Technology Principles Page 12

13 Provider Backbone Bridge Project 802.1ad Interfaces Provider Backbone Bridge Network Provider Bridge Network N Provider Bridge Network Page 13

14 Provider Backbone Bridge Network : Provider Bridge (as defined by 802.1ad) : Backbone Provider Bridge Edge : Backbone Provider Bridge Page 14

15 N Provides Multi-Point B-VLANs Between Ns N N B-VLAN X S-VLAN 4 N S-VLAN 3 N N S-VLAN 1 S-VLAN 2 B-VLAN Y : Backbone Provider Bridge Edge Each B-VLAN carries many S-VLANs S-VLANs may be carried on a subset of a B-VLAN (i.e. all P-P S-VLANs could be carried on a single MP B-VLAN providing connection to all end points. Page 15

16 Provider Backbone Bridge Model Provider Bridge Relays Backbone Bridge Relays MIF 8.5,6.7,9.5 MCF (D 6.5) MAC (802.3) Relay MIF 8.5,6.7,9.5 S-VLAN Map Shim MIF MCF Relay MIF MCF MIF MCF Relay MIF MCF MIF MCF Relay MIF MCF Relay MIF MIF 8.5,6.7, ,6.7,9.5 S-VLAN Map Shim MCF (D 6.5) MAC (802.3) I Imaginary MAC MAC (802.3) MAC (802.3) MAC (802.3) MAC (802.3) Imaginary MAC I Backbone Bridge Interfaces Provider Bridge Interfaces Page 16

17 Backbone Core Relays Can be 802.1ad Provider Bridge Network MIF 8.5,6.7,9.5 MCF (D 6.5) MAC (802.3) Relay MIF 8.5,6.7,9.5 S-VLAN Map Relay MIF MIF MCF MCF Backbone Core MIF Relay MCF MIF MCF Relay MIF MIF MCF MCF Relay MIF MIF 8.5,6.7, ,6.7,9.5 S-VLAN Map MCF (D 6.5) MAC (802.3) Provider Bridge Network I Imaginary MAC MAC (802.3) Backbone Edge MAC (802.3) MAC (802.3) MAC Imaginary MAC (802.3) Backbone Edge I Backbone Core can be single 802.1ad relay Backbone Edge is a dual 802.1ad relay and an encap/decap between the two relays. Page 17

18 Customer,, Spanning Trees Customer Spanning Trees Spanning Trees QB QB QB QB QB QB QB QB QB Spanning Trees QB Spanning Trees Customer spanning trees may extend over Provider Network Network and Network spanning trees must be decoupled to scale the provider network Page 18

19 Functions In Map Shim 802.1ad MIF MCF MAC (802.3) Relay MIF MCF Virtual MAC MIF 8.5,6.7,9.5 S-VLAN Map Shim Backbone Edge 802.1ad MCF (D 6.5) Relay MIF 8.5,6.7,9.5 MCF (D 6.5) MAC (802.3) Does encap/decap of 802.1ad frame Maps S-VID from 802.1ad into larger Extended Service VID (ES-VID) Learns and Correlates Backbone POP and Customer MAC addresses Filters L2 control packets sourced by core relays or by provider bridge relays (divides spanning trees) Page 19

20 S-VLANs Multiplex into B-VLANs B-VLANs S-VLANs Backbone Bridge (802.1ad) I Relay MAP Shim Relay Backbone Provider Edge Bridge I Provider Bridge (802.1ad) MAP Shim performs encap/decap of frames to/from Provider Bridge Networks Page 20

21 Map Shim Encap N encapsulates N frames with N header N header consists of a) Extended Service VLAN identifier Identifies the Provider Bridge S-VLAN within the N Requires 2^20 bits to identify 1M services b) Site Connectivity identifier Identifies a B-VLAN (or tunnel) that is used to transport the N service instance Site connectivity (i.e., tunnel/domain) can be point-to-point or multi-point in nature c) Backbone POP Address MAC Address for POPs within Site Connectivity N Service VLAN IDs (S-VIDs) map to N Extended Service VLAN IDs (ES-VIDs) N S-VIDs are local to the N N ES-VIDs are local to the N Page 21

22 Terminology IEEE 802.1ad Terminology C-TAG Customer VLAN TAG C-VLAN Customer VLAN C-VID Customer VLAN ID S-TAG Service VLAN TAG S-VLAN Service VLAN S-VID Service VLAN ID Additional Backbone Provider Bridge Terminology ES-TAG Extended Service VLAN TAG Field (I-TAG) ES-VID Extended Service VLAN ID (SID) B-MAC Backbone MAC Address B-VLAN Backbone VLAN (tunnel) B-TAG Backbone TAG Field B-VID Backbone VLAN ID (tunnel) Page 22

23 Extended Service VLAN IDs In Backbone S-VLAN 4 S-VID 32 B-VLAN X S-VID 41 ES-VID 4 N S-VLAN 3 S-VID 33 ES-VID 3 S-VID 31 S-VID 42 B-VLAN Y S-VLAN 1 S-VLAN2 : Backbone Provider Bridge Edge An ES-VID uniquely identifies a S-VLAN within the Backbone The MAP Shim translates between S-VID and ES-VID The ES-VID to(from) S-VID mapping is provisioned when a new service instance is created Page 23

24 Single ES-VID per S-VLAN S-VID 2 N S-VID 3 ES-VID S-VID 1 Regardless of the ES-VID address size the map tables only have 4096 entries since only one ES-VID exists per S- VLAN and only 4096 S-VLANs exist per Provider Bridge. A different S-VID in each N maps to the ES-VID Page 24

25 Site Connectivity B-VLAN ID B-VLAN X S-VLAN 4 N S-VLAN 3 S-VLAN 1 S-VLAN 2 B-VLAN Y B-VLANs are addressed like regular VLANs with a 12 bit B-VID B-VID and ES-VID need to be separate ID spaces to allow many S- VLANs to be carried in a single B-VLAN Page 25

26 Backbone POP MAC Address N B-MAC 4 Frame B-MAC 1 Frame DA <- B-MAC 4 SA <- B-MAC 1 Frame B-MAC Addresses identify the Edge Backbone Provider Bridges ( ) B-MAC Addresses are learned by other Edge Backbone Edge Bridges The backbone edge MAC address determines which edge on the B-VLAN will receive the frame. Frames may be flooded by sending with broadcast or multicasts DA B-MACs to the B- VLAN. Map shims filter based on the ES-VID removing any misaddressed frames Page 26

27 Customer/Provider Addresses Provider MAC Addresses MIF MCF Relay MIF MCF Customer MAC Addresses MIF 8.5,6.7,9.5 MCF (D 6.5) S-VLAN Map Shim Relay MIF 8.5,6.7,9.5 MCF (D 6.5) MAC (802.3) Relay Learns Customer Address Per S- VLAN Relay Learns Provider Addresses Per B-VLAN MAP Shim Learns Correlated Customer and Provider MAC Addresses per S-VLAN MAC (802.3) Virtual MAC Backbone Edge Customer/Provider MAC Address Correlation Page 27

28 Backbone Frame Format N Frame Header Frame Format MAC DA MAC SA S-TAG C-tag Payload FCS N Frame Format N Frame Header MAC DA MAC SA S-TAG C-TAG Payload N FCS B-MAC DA B-MAC SA B-TAG ES-TAG Page 28

29 MAP Shim Correlation Table Provisioned S-VID ES-VID B-VID 0x000 0x x0c0 0x001 0x x007 Provider Addresses B-MAC Addresses 0x x Customer Addresses C-MAC Addresses 0x x xfff 0x x0c0 C-MAC Address 0x xdddddddddddd In the beginning the MAP Shim is provisioned with the correlation between the S- VID, ES-VID, and B-VID During operation the MAP Shim learns both B-MAC addresses and C-MAC addresses The MAP Shim keeps track of which C-MAC addresses are behind which B-MAC The correlation data is used to encapsulate frames from the Ns Page 29

30 Basic MAP Shim Operation Frames received from Relay are encaped S-VID is looked up in correlation table to get ES-VID and B-VID C-DA is looked up in C-MAC table to get B-MAC for encapsulation If C-DA is not present in C-MAC table then multicast to B-VLAN Frames received from Relay are de-encaped ES-VID is looked up in correlation table to get a new S-VID B-MAC and C-MAC addresses are learned when frames are received from relay B-MAC and C-MAC addresses are aged Page 30

31 Provider Backbone Bridging 802.1ad Interfaces Provider Backbone Bridge Network (N) Provider Bridge Network N Provider Bridge Network Page 31

32 Terminology IEEE 802.1ad Terminology C-TAG Customer VLAN TAG C-VLAN Customer VLAN C-VID Customer VLAN ID S-TAG Service VLAN TAG S-VLAN Service VLAN S-VID Service VLAN ID Additional Backbone Provider Bridge Terminology ES-VID Extended Service VLAN ID B-VLAN Backbone VLAN (tunnel) B-VID Backbone VLAN ID (tunnel) B-MCD Backbone Multicast Domain B-TAG Backbone TAG Field B-MAC Backbone MAC Address Page 32

33 A Simple Provider Backbone Bridge Network : Provider Bridge (as defined by 802.1ad) : Provider Backbone Bridge Edge : Provider Backbone Bridge Page 33

34 Simple N Principles edge encapsulates received N frames with N header N header includes Extended Service VLAN Identifier (ES-VID) Identifies the S-VLAN associated with the N S-VIDs on the N Must be large enough to support millions of S-VLANs Backbone VLAN Identifier I (B-VID) Identifies a backbone VLAN (B-VLAN or tunnel) that is used to transport the S-VLANs over the N A B-VLAN(tunnel) can be point-to-point or multi-point in nature The B-VID must have a large enough address space to support all available multi-point tunnels among bridges Backbone POP Address Addresses POP within Site Connectivity Page 34

35 Use Hierarchical Architecture to Scale N size Support of B-VLANs (i.e., multicast) with large number of bridges is challenging frame replicators to large number of points limit performance Hierarchy of bridges creates small multicast domains each domain has a small number of bridges, which limits number of multi-point tunnels and number of replications Page 35

36 A Two Layer Hierarchical N Layer 1 MC Domain A Layer 1 MC Domain B Layer 2 MC Domain A Layer 1 MC Domain C Layer 1 MC Domain D : Provider Bridge (as defined by 802.1ad) : Backbone Provider Bridge Edge : Backbone Provider Bridge : Backbone Provider Bridge Layer Edge Page 36

37 Hierarchical N Principles edge Encapsulates received N frame with N header swaps S-VID to/from a much large ES-VID Creates a B-VID from the ES-VID De-encapsulates frames to be transmitted to the N by stripping the N header Swaps the ES-VID to a S-VID for the N Removes the final B-VID Both S-VID and ES-VID identify the S-VLAN carried through the Ns and N layer edge bridge swaps the B-VID to a new B-VID based on the ES-VID The new B-VID allow transport over the current N multicast domain (MC-DOM) source route addressing (with Backbone Connectivity identifier stacking) can also be used to avoid the need for table lookups and B-VID swapping at layer boundaries The ES-VID is the same throughout the N The ES-VID is swapped with the S-VID at the - edge B-VID must be large enough to address all possible multi-point tunnels within a given layer domain (e.g., 12 bits is enough to support 12 bridges in a layer. More bits are required for more bridges). Scalability Hierarchical N can have as many layers as required Page 37

38 Summary A Provider Backbone Bridge standard will allow highly scalable Ethernet Backbone The 802.1ad control plane may be used on both sides of the MAP Shim Connection Fault Management 802.1ag supported by the Provider Backbone Bridges Page 38

39 MEF Services and Specifications Page 39

40 Agenda MEF Specifications Roadmap Services Model Traffic Management Page 40

41 Technical Committee Organization Technical Committee Services Architecture Protocol & Transport Management Test* Phase I Clean-up Reference Model Protection EMS Phase II Extensions PDH Circuit Emulation User Network Interface Inter-carrier Interface Multiplexing Ethernet over SONET Information Model OA&M Service Test Sets At UNI SONET/SDH Circuit Emulation Page 41

42 9 MEF Specifications MEF 1 MEF 2 MEF 3 MEF 4 MEF 5 MEF 6 MEF 7 MEF 8 MEF 9 Ethernet Service Model TS Requirements and Framework for Ethernet Service Protection in MENs TS CES Definitions, Framework and Requirements in MENs TS Metro Ethernet Architecture Framework- Part 1: Generic Framework TS Traffic Management TS Ethernet Service Definitions TS EMS-NMS Information Model TS Implementation Agreement for PDH Emulation Service IA Abstract Test Methods For Ethernet Service at UNI Available on the MEF public WEB at Page 42

43 MEF Services Technical Specifications MEF 1 Ethernet Services Model, Phase1* Technical descriptions of service features MEF 5 Traffic Management Specification, Phase1 Fractional Bandwidth and Performance MEF 6 Ethernet Services Definitions, Phase1 Specific service instances * Page 43

44 Agenda MEF Specifications Roadmap Services Model Traffic Management Page 44

45 Services Model Customer Edge (e.g., router) (CE) Customer Edge (CE) Metro Ethernet Network Service Attributes A service is what the CE sees. The technology used inside the MEN is not visible. Page 45

46 User Network Interface The demarcation point between Service Provider and Subscriber Responsibilities CE UNI Metro Ethernet Network UNI CE Dedicated to a single Subscriber Based on Standard Ethernet PHYs for Phase 1, e.g., RJ45 Socket on Service Provider owned Ethernet switch RJ45 plug on Service Provider owned cable Page 46

47 Service Frame The Layer 2 protocol data unit exchanged between the CE and the MEN at the UNI Standard Ethernet With IEEE 802.1Q tag (up to 1522 bytes) Without IEEE 802.1Q tag (up to 1518 bytes) Includes everything but the preamble More than 100 Million devices exist that are potential Customer Edge devices Page 47

48 Service Frame Transparency Service Frames must be delivered from ingress UNI to egress UNI(s) transparently except possibly as follows: Ingress Service Frame Untagged Tagged Tagged Egress Service Frame* Tagged Untagged Tagged w/ different value *Frame Check Sequence recalculated Page 48

49 Each Service Instance is a Layer 2 VPN Example showing a green service and a blue service. Service Multiplexed UNI Service Frames cannot leak in or out of a Service Instance Multiple Service instances can exist at a UNI, called Service Multiplexing Page 49

50 Formal Service Instance Definition Ethernet Virtual Connection (EVC) Association of two or more UNIs Service Frames can only be exchanged among the associated UNIs A Service Frame sent into the MEN via a particular UNI MUST NOT be delivered out of the MEN via that UNI Page 50

51 Point-to-Point EVC Exactly two UNIs are associated. Page 51

52 Multipoint-to-Multipoint EVC Two* or more UNIs are associated A broadcast or multicast ingress frame is typically replicated and delivered to all of the other UNIs * A MP2MP EVC with two UNIs is different than a P2P EVC since additional ditional UNIs can be added at any time. Page 52

53 Identifying an EVC at a UNI CE-VLAN ID/EVC Map Service Frame Format Untagged* Priority Tagged* Tagged, VID = 1 Tagged, VID = 2... Tagged, VID = 4094 Tagged, VID = 4095 CE-VLAN ID CE-VLAN ID/EVC Map EVC Red Green... Blue *Untagged and Priority Tagged Service Frames have the same CE-VLAN ID and that value is configurable at each UNI. This is the behavior expected by an IEEE 802.1Q CE. Page 53

54 CE-VLAN ID Preservation CE-VLAN ID 37 EVC Blue EVC Blue CE-VLAN ID 37 CE-VLAN ID/EVC Map for EVC must be identical at all UNIs in the EVC and Priority Tagged in must be priority tagged out Untagged in must be untagged out Page 54

55 All to One Bundling (Map) Untagged* Priority Tagged* Tagged, VID = 1 Tagged, VID = 2... Tagged, VID = 4094 Tagged, VID = 4095 CE-VLAN ID CE-VLAN ID/EVC Map EVC Red Only one EVC at the UNI (no service multiplexing) All CE-VLAN IDs map to this EVC no need for coordination of CE-VLAN ID/EVC Map between Subscriber and Service Provider EVC must have CE-VLAN ID Preservation Page 55

56 Using All to One Bundling Disaster Recovery Service Provider Branch HQ Branch Bridge or Router Private Line Replacement LAN Extension Page 56

57 Untagged Priority Tagged Tagged, VID = 1 Tagged, VID = 2... Tagged, VID = 4094 Tagged, VID = 4095 One to One Map CE-VLAN ID CE-VLAN ID/EVC Map EVC Red Blue No more than one CE-VLAN ID is mapped to each EVC at the UNI If CE-VLAN ID not mapped to EVC, ingress Service Frames with that CE- VLAN ID are discarded Service Multiplexing possible CE-VLAN ID Preservation not required Subscriber and Service Provider must coordinate CE-VLAN ID/EVC Map Page 57

58 CE-VLAN ID Translation CE-VLAN ID 37 EVC Blue EVC Blue CE-VLAN ID 156 CE-VLAN ID/EVC Map can be different at different UNIs in an EVC Fine for CE routers Problematic for CE bridges Page 58

59 Untagged* Priority Tagged* Tagged, VID = 1 Tagged, VID = 2... Tagged, VID = 4094 Tagged, VID = 4095 Bundling (Map) CE-VLAN ID CE-VLAN ID/EVC Map More than one CE-VLAN ID is mapped to an EVC at the UNI Service Multiplexing possible EVC Red Blue CE-VLAN ID Preservation is required for EVC if multiple CE-VLAN IDs mapped to it Subscriber and Service Provider must coordinate CE-VLAN ID/EVC Map Page 59

60 Feature Combinations and Uses CE-VLAN ID/EVC Map Characteristic Point-to-Point EVC Type Multipoint-to-Multipoint All to One Bundling One to One Map Bundling Private Line replacement with Router or Bridge Frame Relay replacement with Router Uses TBD Ethernet Line Service (E-Line) LAN Extension with Router or Bridge Uses TBD Uses TBD Ethernet LAN Service (E-LAN) Page 60

61 Delivery of Service Frames Broadcast Deliver to all UNIs in the EVC but the ingress UNI Multicast Typically delivered to all UNIs in the EVC but the ingress UNI Unicast Typically delivered to all UNIs in the EVC but the ingress UNI if not learned Otherwise, deliver to the UNI learned for the destination MAC address Learning is important for Multipoint-to-Multipoint EVCs Type of Service Frame determined from the destination MAC address Page 61

62 Options for Layer 2 Control Discard Protocols PDU from CE discarded by MEN PDU never egresses from MEN Peer MEN peers with CE to run protocol Tunnel PDUs carried across MEN as if they were normal data EVC is that associated with the CE-VLAN ID of the PDU, e.g., the Untagged CE-VLAN ID for most standard Layer 2 Control Protocols defined by IEEE 802 Page 62

63 Agenda Specifications Roadmap Services Model Traffic Management Page 63

64 Two Areas Covered by Traffic Bandwidth Profile Management How to buy just the bandwidth you need and have a predictable bill Class of Service Identifying the CoS for a Service Frame Performance parameters that define a CoS Page 64

65 Bandwidth Profile Overview Similar in concept to the traffic policing of Frame Relay Bandwidth Profile is a characterization of the lengths and arrival times of Service Frames at the UNI The level of compliance with the Bandwidth Profile is assessed for each ingress Service Frame Green = full compliance Yellow = partial compliance Red = non-compliance Delivery performance then based on compliance level Page 65

66 Bandwidth Profile Defined by Token Bucket Algorithm Green Tokens Committed Information Rate Yellow Tokens Excess Information Rate Committed Burst Size Overflow Excess Burst Size Overflow C-Bucket E-Bucket If (Service Frame length less than C-Bucket C tokens) declare green and remove tokens from C-BucketC else if (Service Frame length less than E-Bucket E tokens) declare yellow and remove tokens from E-BucketE else declare red Page 66

67 Two Options for Algorithm Option 1: Decoupled Option 2: Coupled Green Tokens Green Tokens C-Bucket Overflow C-Bucket Overflow Yellow Tokens Overflow Yellow Tokens Overflow E-Bucket E-Bucket Page 67

68 Bandwidth Profile Parameters Committed Information Rate (CIR) expressed as bits per second. CIR 0. Committed Burst Size (CBS) expressed as bytes. CBS 0. Excess Information Rate (EIR) expressed as bits per second. EIR 0 Excess Burst Size (EBS) expressed as bytes. EBS 0. Coupling Flag (S). S = 0 or 1. Page 68

69 Three Ways to Apply Bandwidth Profile to a Service Frame Per ingress UNI Per EVC at the ingress UNI Per CoS instance at the ingress UNI (see below for CoS identification) Multiple methods can apply at a UNI but configuration must be such that only one Bandwidth Profile is applied to each ingress Service Frame Page 69

70 Bandwidth Profile Policing Green Deliver Service Frame with performance levels as per the Service Level Agreement for the CoS instance Yellow Deliver Service Frame but Service Level Agreement for the CoS instance does not apply Red Discard Page 70

71 EVC Two Ways to Identify CoS Instance All Service Frames mapped to the same EVC receive the same CoS <EVC,set of user_priority values> All Service Frames mapped to an EVC with one of a set of user_priority values receive the same CoS Page 71

72 Class of Service A Class of Service is defined by three performance objectives Frame Delay: P percentile of delay d msec Frame Jitter: Definition TBD Frame Loss: Percent of frames lost p% Phase 1 will cover only Point-to-Point EVCs Page 72