1、Bob Metcalfes “Ethernet“,Fig. 1 A drawing of the first Ethernet system - from Bob Metcalfe (, 3),Ethernet - Frame structure,Preamble 前導碼,SFD 幀起始定界符,Destination 目的地址,Source 源地址,Type field 類型,Data Field 數(shù)據(jù),FCS 幀校驗序列,7 Byte,1 Byte,6 Byte,6 Byte,2 Byte,Variable Length,64 Byte Frame Size 1518 Byte 1519 B

2、yte Jumbo Frame 9018 Byte,4 Byte,Fig. 2 Ethernet Frame Structure (, 7),Ethernet - MAC-Address,6 Octets,12 hex. digits,48 bits,1,2,3,4,5,6,Manufacturer,Equipment,1,2,3,4,5,6,7,8,9,10,11,12,Fig. 3 Ethernet MAC Address (, 7),Traffic Control (Policing & Shaping) Ethernet Priorization (IEEE 802.1p, etc.)

3、 Virtual LAN (IEEE 802.1Q) Spanning Tree Protocol (IEEE 802.1D) Ethernet Ring Protection (ERP) Link Aggregation Group (IEEE 802.3 ad) Ethernet First Mile (IEEE 802.3 ah) Multicast (IGMP) Routing Protocols Additional Features (Jumbo Frame Support, MAC-Address Translation, DHCP Relay Agent, ) Multi Pr

4、otocol Label Switching - MPLS,Technologies used in Carrier Ethernet,Fig. 4 Protocols and Technologies (, 9),IP Multiservice Network,E-mail Servers,Mainframe,PC,PC,Telephone,PBX,CoS is describing a set of features and other characteristics available with a specific service or service package,Best Eff

5、ort,Standard,Premium,Class of Service,What is Class of Service (CoS)?,Fig. 5 What is COS - Class of Service? (, 11),Real-Time Flow,Router,Router,Workgroup Switch,Remote Switch,Data Server,VoIP Server,VoIP Server,Data Server,IP Multiservice Network,QoS is a more precise term, used to measure a specif

6、ied set of performance attributes typically associated with a service. IP QoS refers to the performance of IP packets flowing through one or more networks.,Non-Real-Time Flow,Non-real-time service flows may not require specified throughput, delay, delay variation, packet loss and availability parame

7、ters,Real-time service flows require specified throughput, delay, delay variation, packet loss and availability parameters,What is Quality of Service (QoS) ?,Fig. 6: What is QoS - Quality of Service? (, 11),Basic Concepts to handle User Traffic,Packet,IncomingTraffic,Packet,Packet,ForwardedTraffic,P

8、acket Filtering,Packet Queuing,Packet Fragmenting,Packet Forwarding,Packet Dropping,Fig. 7: Basic Concepts to handle User Traffic (, 13),Precedence Priority,Queue Priority,Packet Dropping,IncomingTraffic,ForwardedTraffic,high,low,middle,Packet Filtering and Precedence Prioritization,Packet Queuing,F

9、ig. 8: Precedence Priority (, 15),Selective Packet Discard,ForwardedTraffic,IncomingTraffic,Congestion Threshold,Packet Dropping,Packet Filteringand Marking,Packet Queuing,Fig. 9: Selective Packet Discard (, 15),Traffic Descriptors,CIR (Committed Information Rate)- guaranteed minimum rate- configure

10、d per logical (802.1Q) port PIR (Peak Information Rate)- maximum rate packets are allowed- PIR CIR- configured per logical (802.1Q) port MBS (Maximum Burst Size) - MBS defines together with the PIR the maximum network load without discard,Fig. 10 Traffic Descriptors (, 17),Leaky Bucket Algorithm,Inc

11、oming traffic of a FC on a logical port,Non-conforming Ethernet Frames can be marked or dropped,Conforming Ethernet Frames are transmitted,Traffic at or below CIR is considered as compliant. Traffic over CIR is marked as out-of-profile“ or dropped.,Hole Size corresponds to CIR,Fig. 11 Leaky Bucket A

12、lgorithm (, 19),Traffic Shaping,Customer Equipment usually burst with line rate with any regard to PIR Ingress Shaping ensures that service traffic is accepted into the Service Provider network at the PIR Ingress Shaping provides better service by absorbing the traffic burst up to MBS without disadv

13、antage of hard“ policing (dropping or marking) BUT: Traffic over PIR and MBS is dropped,Intended data-rate (e.g. CIR),per connection buffer,Incoming data from CPE,Shaping,Outgoing data (shaped),Fig. 12 Shaping (, 19),Incoming Queues,Forwarder,Outgoing Priority Queues,Classification,1,1,2,2,3,3,High,

14、Medium,Normal,4,Low,Strict Priority (SP) Queuing,Fig. 13 Strict Priority Queuing (PQ) (, 21),Weighted Random Early Discard (WRED),Packet Discard Probability,Queue Utilization,Min. Threshold,Max. Threshold,0,1,PHB AF1i Drop Probability,PHB AF Per Hop Behavior for Assured Forwarding Traffic Class,Fig.

15、 14 WRED (, 23),Weighted Fair Queuing (WFQ),Number of Queues configurable,Incoming packets,Weighted Fair Scheduling,Outgoing packets,Classify(flow-based),Fig. 15 Weighted Fair Queuing (WFQ) (, 25),Weighted Round Robin (WRR),Number of Queues configurable,Incoming packets,Weighted Fair Scheduling,Outg

16、oing packets,Classify(flow-based),Fig. 16 Weighted Fair Queuing (WFQ) (, 25),L2 - Ethernet Priorization - IEEE 802.1p,Dest,Src,LEN/Type,Data,FCS,Dest,Src,LEN/Type,Data,FCS,Etype0 x8100,TAG,3 Bit,VLAN ID,Ethernet-Frame: 802.3,Ethernet-Frame: 802.1Q,2 Bytes,PRI,Fig. 17 Ethernet Priorization IEEE 802.1

17、p (, 27),L3 Priorization - The IP Header,IP Version,Length of the IP packet,Length of the IP header,Identification of the IP packet (used together with the flags for Reassembly),Lifetime of the IP packet,Type of the higher protocol,Additional, optional Information,TOS = Type of Service,Fig. 18 Struc

18、ture of the IP header (, 27),3 priority bits,traffic classes,outgoing queues(max. 8),IEEE 802.1p,Fig. 19 IEEE 802.1p (, 29),Priority Bit Setting and Queuing,If a switch provides 8 queues:,If a switch provides less than 8 queues:,each priority is mapped to one traffic class/queue,priorities has to sh

19、are traffic classes/queues,Fig. 20 Priority Bit Setting and Queuing (, 29),VLAN Virtual LAN,Physical View,Logical View,Users,Switches,Router,Marketing LAN,Engineering LAN,Manufacturing LAN,Fig. 21 VLAN Physical and Logical View (, 31),Virtual LAN (VLAN) - IEEE 802.1Q,Dest,Src,LEN/Type,Data,FCS,Dest,

20、Src,LEN/Type,Data,FCS,Etype0 x8100,TAG,PRI,12 Bit,VLAN ID,Ethernet-Frame: 802.3,Ethernet-Frame: 802.1Q,2 Bytes,Fig. 22 VLAN 802.1Q (, 33),VLAN Stacking (QinQ“),Users,Customer VLAN,Provider VLAN,Inner Tag (Customer VLAN ID),Outer Tag (Provider VLAN ID),Users,Fig. 23 VLAN Stacking “QinQ” (, 33),Spanni

21、ng Tree 802.1d,redundant topology,redundant free topology,Loops,STP back-up mode,Root Bridge,without STP,with STP,DesignatedRoot Bridge,Bridge Protocol Data Unit (BPDU),Fig. 24 Spanning Tree 802.1d (, 35),Multiple Spanning Tree (802.1s),physical link or LAG,tree for the red group of VLANs,tree for t

22、he green group of VLANs,combines STP with VLAN (normal STP doent support VLAN) allows for each VLAN to calculate its own spanning tree supports multiple links (LAG) usefull for load distribution,Fig. 25 Multiple Spanning Tree (802.1s) (, 37),Ethernet Ring Protection ERP,ERPMaster,ERP master blocks t

23、raffic of payload VLANs,Payload VLAN (user data),Control VLAN,Payload (Protected) VLANs for user data 1 (!) Control VLAN for monitoring and control Each Payload VLAN is assigned to the Control VLAN Control VLAN managed by ERP master,Fig. 26 Ethernet Ring Protection (ERP) (, 39),ERP Failure Detection

24、 and Recovery,Master ERP Switch,Fast detection: Link failure reportfrom ERP-aware switches Slow detection: Loopback packets for ring integrity (on Control VLAN),Link failure detection:,! Link down,! Link down,Ring reconfiguration,Master ERP Switch,Recovery:,Fast recovery: - Immediate flush of forwar

25、ding tables (initiated by Control VLAN)- release of blocked ports Reconfiguration: - 50ms fail-over,Ring reconfiguration,Fig. 27 ERP Failure Detection and Recovery (, 39),Link Aggregation Group LAG (802.3ad),a group of physical ports creates a logical port requirement: all ports of a group must supp

26、ort- same speed- same duplex-capability (full-duplex) provides:- fault protection- higher bandwidth LACP (Link Aggregation Control Protocol)- monitoring of LAG,Fig. 28 Link Aggregation Group LAG (802.3ad) (, 41),Video Broadcast and IGMP,IP/MPLS,Ethernet,CPE,TV + STB,ERX,BTV head-end,Video Server + E

27、PG,ISP 1,ISP 2,Ethernet,hiD,DSLAM,IGMPJoin,IGMPJoin,IGMP Router,IGMP Host,IGMP Proxy,Fig. 29 SDVB & IGMP (, 43),IGMP Messages,Which group members? IGMP membership query,Well-Known Multicast Adresses,Fig. 30 IGMP Messages (, 43),IGMP Snooping,IP/MPLS,Voice,ISPs,ASPs,DSLAM/Switch is reading all data f

28、rames and interprets IGMP messages By evaluating IGMP joins and IGMP query reports, the DSLAM can recognize which multicast group can be reached via which DSL port,IGMP messages,listening to the IGMP conversation between router and hosts,Fig. 31 IGMP Snooping (, 45),IGMP Proxy,hiX DSLAM,Fig. 32 IGMP

29、 Proxy (, 45),Interior and Exterior Routing,An Autonomous System (AS) is a collection of routers with a single coherent interior routing plan Interior routing protocol (IGP) is run inside an autonomous system resulting in optimum intra-AS routing Exterior routing protocol (EGP) is run between autono

30、mous systems to enable routing policies,EGP,IGP,AS 1,AS 2,IGP,BGP,RIP/OSPF,RIP/OSPF,Fig. 33 Interior and Exterior Routing (, 47),Routing Information Protocol (RIP),Interior Routing Protocol suitable for small-seized networks RIP uses the Distance Vector Algorithm Hop-Count is being used as metric (m

31、ax. number of hops: 16) RIP messages are being distributed via broadcasts RIP but doesnt scale very well in large enterprise networks Routing loops may occur RIP has slow convergence times (n * 30 sec) RIP Version 2 has been developed to overcome the limitations of version1 SUMMARY: - RIP is a low-b

32、rainer“ - RIP is suitable for small networks,Fig. 34 RIP (, 49),Open Shortest Path First (OSPF),Interior Routing Protocol suitable for medium-seized and larger networks fast convergence time (in comparison to RIP) supports VLSM (Variable Length Subnet Mask) metric based on link bandwidth efficient u

33、se of the bandwidth for updates supports several routes to the destination (redundancy) allows hierarchical networks (based on areas) SUMMARY: - OSPF scales better than RIP - but some administration work to do (area config),Fig. 35 OSPF (, 49),Border Gateway Protocol (BGP),Exterior Routing Protocol

34、BGP allows the definition of policies or rules, describing the flow of data through the Autonomous System Distance-vector protocol with enhancements (comparing to RIP): - policy-based routing- reliable updates (TCP port 179)- incremental updates and keepalives- rich set of metrics (path attributes)-

35、 designed to scale in huge internetworks- internet has today over 100.000 routes and is still growing SUMMARY: - very powerful, but complex to administrate- suitable for hugh routing tables,Fig. 36 BGP (, 51),Jumbo Frames,will be supported from hiX5630 (from Rel. 1.5 on) not a fixed term according t

36、o MTU (max. Transmission Unit) means every supported frame length larger than 1,500 bytes survey in 2003:- 65% at 4,470 bytes- 8% between 9,000 - 9,176 bytes- 27% at 9,180 bytes and above.,Normal“ Ethernet Frame:,Max. 1,500 bytes (no header and CRC!),Jumbo Ethernet Frame:,Max. 9,xxx bytes (no header

37、 and CRC!),min.,Fig. 37 Ethernet Jumbo Frames (, 53),MAC Address Translation,users MAC address will be replaced by MAC address of DSLAM advantage: security (protection against DoS attacks),IP/MPLS,Ethernet,DSLAM(MAC-:D),CPE,Router R(MAC-:R),Video Server,ISP 1,ISP 2,Ethernet,hiD,User A(MAC-:A),provid

38、esMAC- translation,SA: Source MAC-, DA: Destination MAC-,Fig. 38 MAC address translation (, 55),Dynamic Host Configuration Protocol (DHCP),to obtain an IP-adress (DHCP Query) additionally:- Subnetwork-Mask- Default Gateway Router- DNS-Server BUT: DHCP-Query is a local request (per subnetwork!) Solution: Configuration of a DHCP Relay Agent - forwards DHCP-Request to a DCHP-Server outside of the subnetwork,DHCP Request,DHCP Client,DHCPRelay Agent,(broadcast),DHCP Request,(unicast),DHCPServer,Fig. 39 DHCP (, 5