IP Traffic Engineering and QoS

Slides:



Advertisements
Similar presentations
Copyright© 2005 DASAN Networks, Inc. V61XXF NOV
Advertisements

1 2 장. 데이터 통신. 2 데이터 통신망 형태  점대점 (point-to-point)  통신망 (communication network)  전화망 (PSTN: Public Switched Telephone Network)  LAN(Local Area Network)
GMPLS Generalized Multi-Protocol Label Switching.
1 IP Multicasting ♠ 인터넷 전송 방식 종류와 차이점 - 유니캐스트 - 브로드캐스트 - 멀티개스트 그림 1. 브로드캐스팅의 경우 NON-Multicast 응용 UDP IP MAC Multicast 응용 UDP IP MAC Multicast 응용 UDP IP.
멀티미디어 서비스를 위한 IP 네트워크 순천향대학교 정보기술공학부 이 상 정
Mobile IPv6 (RFC 3775) 윤창식 황동윤.
Building Enterprise VPNs
Differentiated Services
Chapter 7 ARP and RARP.
TCP/IP 통신망 특론 3장 인터넷프로토콜( IP ).
Internet Protocol Version4
Mobile IPv 여지민 송구득 박근홍 (수)
Routing.
컴퓨터 네트워크 Chapter 5-2 컴퓨터 네트워크.
Next Generation: IPv6 and ICMPv6
암호화 기술(SSL, IPSec) 손재성 권기읍 안복선 최준혁
Battery-Dynamics Driven TDMA MAC Protocols for Wireless Body-Area Monitoring Networks in Healthcare Applications Hang Su, Student Member, IEEE, and Xi.
IPsec 석진선.
이번 시간에는... 지난 시간에는 모바일 응용 프로토콜의 첫번째 시간으로, 모바일에 사용되는 다양한 응용 프로토콜과 모바일 웹 서비스 응용 프로토콜로써 XML 및 SMIL 동기화 멀티미디어 통합 언어에 대해 알아보았습니다. 이번 시간에는 모바일 응용 프로토콜, 그 두번째.
UDP 1434 공격에 대한 방어 요령 Cisco Systems Korea 최 우 형 Network 보안과 대책
Networking and Internetworking Devices
Switching 기술 II(L4, L5, L7).
MPLS Architecture and Applications
Computer Communications LAB.,
IPv6 도입과 동향.
20 장 네트워크층 프로토콜 : ARP, IPv4, ICMP, IPv6, ICMPv6
Internet Group Management Protocol (IGMP)
Delivery and Routing of IP Packets
Internet Control Message Protocol (ICMP)
유비쿼터스 네트워크 연구실 (Ubiquitous Network Lab) Since 1994
LOGO 네트워크 운용(2).
IGRP(Interior Gateway Routing Protocol)
PART 01 개념 컴퓨터 네트워크 Chapter 3 OSI 참조모델과 인터넷 임효택.
라우팅의 기술 RIP과 OSPF의 개요 및 동작과정 1조 : 박지훈, 최정연, 추태영 RIP과 OSPF의 개요 및 동작과정.
Chapter 2 OSI 모델과 TCP/IP 프로토콜.
General VoIP 발표자: 황인욱.
(Next Generation Internet Protocol)
SK Telecom Platform연구원
Chapter 11 Unicast Routing Protocols.
라우팅 기술 (RIP, OSPF) 컴퓨터공학과 강지훈 윤인선 이고운
7장 목차 7.1 멀티미디어 네트워킹 응용 7.5 다양한 서비스 클래스 제공 7.2 스트리밍 저장 오디오 및 비디오
Routing Protocol - Router의 주 목적 중 하나는 Routing
Chapter 8 교환 (Switching).
Chapter 8 교환 (Switching).
Routing Enhancements for GMPLS
Dept. Computer science of Engineering
Carlos Guimar˜aes1, Daniel Corujo2, Rui L. Aguiar3
IPSec (Internet Protocol Security protocol)
제 4 장 IP 주소지정 진 표기법 4.2 클래스 4.3 특수 주소 4.4 예제 인터넷
PART 01 개념 컴퓨터 네트워크 Chapter 3 OSI 참조모델과 인터넷 임효택.
2019년 1월 2일 인터넷 QoS 관리 기술 2019년 1월 2일
TCP/IP 통신망 특론 2장 Link Layer 컴퓨터 네트워크 실험실 이희규.
RPL: Routing Protocol for Low Power and Lossy Networks
Internet QoS Technique
IP(Internet Protocol)
Monday_10.29 라우팅 정리 Routing 패킷에 대한 목적지 IP주소와 일치하는 경로를
IPv 이 동 주 HONGIK UNIVERSITY.
Inje Univ. Network Team2 Jung Chang Hwan
박 태하 ㈜ 아이네트 인터넷 망관리를 위한 도구 박 태하 ㈜ 아이네트.
5과목 데이터통신 강사 이 민 욱.
5장. 네트워크 계층과 라우팅 내용 네트워크 계층(Network Layer) 소개 라우팅 알고리즘 IP 프로토콜
Chapter 4 네트워크 계층 소개.
Internet Protocol Objectives Chapter 8
Advances in Optical Technology & FI
Internet & WWW Protocols
QoS Routing in the Internet
ATM - 다양한 종류의 Service 가능 - LAN 과 WAN 에 모두 적합하다. - 전세계적으로 인정된 표준
제 9 장 ICMP 9.1 메시지 유형 9.2 메시지 형식 9.3 오류 보고 9.4 질의 9.5 검사합 9.6 ICMP 설계
Dynamic Graph Query Primitives for SDN-based Cloud Network Management Ramya Raghavendra, Jorge Lobo, Kang-Won Lee 2012 HotSDN 정보통신공학과.
Chapter 7: Deadlocks.
Presentation transcript:

IP Traffic Engineering and QoS 2001년 2월 14일 명지대학교 홍 석 원 서울 시립대 안 상 현

목차 IP Traffic Engineering IP QoS Traffic Engineering 구성 요소 경로 설정과 Constraint-based routing TE signaling과 MPLS IP QoS Motivation Approach Intserv RSVP Diffserv Reference

Traffic Engineering “A major goal of Internet Traffic Engineering is to facilitate efficient and reliable network operations while simultaneously optimizing network resource utilization and performance.” (RFC 2702 Requirement for Traffic Engineering over MPLS) Intra-domain unicast traffic

IP Traffic Engineering 목표 Efficient network를 위해서, “constraint”를 고려한 경로 설정과 트래픽 할당 load balancing과 congestion 해소 서비스의 질을 높임, 등등 Reliable network를 위해서, 링크/노드 고장 시 이를 검출하고, 대체 경로 설정 트래픽의 rerouting 결국 망 보호/복구

IP Traffic Engineering의 접근 Interior Gateway Protocol(IGP) constraint 기반의 routing을 하지 못한다. Layer 2 switching 망의 virtual circuit을 이용 Constraint-based routing을 이용 MPLS의 신호 프로토콜 사용

MPLS TE 요소 경로 설정 트래픽 할당 MPLS에 의한 패킷 전달 경로 관리 경로 정보 관리 link resource 정보 관리 망 상태 정보의 전달: IGP의 Link state packet의 확장 경로 선택 signaling 트래픽 할당 traffic을 경로에 할당 MPLS에 의한 패킷 전달 경로 관리 트래픽을 monitor/measure 망 보호/복구

경로 설정 경로 정보 관리 link resource 정보 관리 경로 계산 경로 설정을 위한 signaling Traffic trunk Traffic trunk attributes resource attributes link resource 정보 관리 정보의 전달(distribution) 경로 계산 경로 설정을 위한 signaling

Traffic Trunk An abstraction of aggregated traffic flow that follows the same path(s) (within a service provider) between two end points 예로서 a POP to another POP 동일한 QoS 요구 사항을 갖는다. LSP tunnel과 동의어로 사용될 수 있다. ATM의 VC와 유사한 개념 Traffic trunk R R R R R R R

Traffic Trunk attributes 경로 설정의 과제 Traffic trunk Traffic Trunk attributes R R source destination Physical network Resource attributes R R R R R R R

Traffic Trunk Attributes(1) Traffic parameter attributes 트래픽 특성: peak, average rates, burst size 등 결국, resource requirements(bandwidth) priority attribute setup priority holding priority preemption attribute resilience attribute 장애가 발생했을 때 trunk를 어떻게 조치해야 하는가를 규정한다.

Traffic Trunk Attributes(2) resource class affinity attributes 특정 resource class(color)를 포함할 것인지를 결정한다. Affinity attribute는 include/exclude의 두 값을 갖는다. 따라서 <resource class, affinity>의 tuple로서 주어진다. administratively specified explicit paths adaptivity attribute 망의 변화에 따라 re-optimization을 할 것인지를 규정한다.

Resource Attributes 모든 link는 resource attributes를 갖는다. Bandwidth link attributes TE-specific link metrics 등 resource attributes 정보는 모든 라우터들에게 broadcast된다. IGP의 LSA로서 전달 따라서 IGP의 LSA의 확장 필요

Trunk와 link resource attributes예 Priority=3 bandwidth= 50 Traffic trunk R R source destination Resource attributes Physical network BW(3)=80 R BW(3)=100 R BW(3)=20 60 100 R R BW(3)=80 R 70 50 BW(3)=40 BW(3)=30 R R BW(3)=100

Resource Class Affinity 망 관리자에 의해서 각 link에 설정된 policy를 의미한다. Affinity 각 link는 특정 policy attribute를 포함할 수도 있고 무시할 수도 있다. 즉 망의 policy에 의해서 결정된다. Affinity는 binary variable로서 특정 policy attribute를 포함(1) 혹은 배제(0)를 명시하는 값을 나타낸다.

Resource Class Affinity 예 예를 들면, 네 개의 policy attribute가 존재하고 세번째 attribute가 restoration 가능 여부를 나타내는 attribute라고 하자. 그러면 각 link는 다음의 bit string으로 resource class affinity를 나타낼 수 있다. LSP tunnel의 attribute가 0010라면, 빨간 색 link 만이 가능하다. 0000 R 0010 R 0000 0010 0100 R R 0010 R 0010 0010 0010 0010 R R 0000

경로 선택 constrained-based routing에 의한 경로 선택 Input(constraint) 결과 traffic trunk attributes link resource attributes topology state information 결과 constraint를 만족하는 explicit path(a list of routers)

경로 계산 Constraint를 만족하지 못하는 link는 잘라낸다.(prune) 나머지 link들 중에서 shortest path를 구한다. TE-specific metric을 사용한다.

앞의 예에서, Traffic trunk Priority=3 bandwidth= 50 tunnel attribute=0010 R BW(3)=80 0000 R BW(3)=100 R BW(3)=20 0010 0010 0000 BW(3)=60 0100 BW(3)=80 R BW(3)=100 R R 0010 0010 0010 0010 BW(3)=50 BW(3)=70 0010 BW(3)=30 R BW(3)=40 R 0000 BW(3)=100

경로 설정 signaling MPLS의 signaling protocol을 사용하여 Explicit route(ER)을 설정한다. CR-LDP (constrained-Based LSP Setup using LDP) RSVP-TE (RSVP Extension for LSP)

CR-LDP로 ER-LSP 설정 A B C D Explicit Route LSR LSR LSR LSR Label Request message 전송: ER path <B,C,D> Label Request message 처리: ER path <C,D> Label Request message 처리: ER path <D> Label Mapping message 전송

RSVP-TE로 ER-LSP 설정 A B C D Explicit Route LSR LSR LSR LSR Path message 전송 : ERO <B,C,D> RRO <A> Path message 처리: ERO <C,D> RRO <A,B> Path message 처리: ERO <D> RRO<A,B,C> RESV 전송

망 상태 정보의 전달 망 노드 간에 topology state 정보를 전달한다. 이것은 OSPF의 Link State의 전달과 동일하다. 트래픽 엔지니어링을 위해서 현재 IGP Link State Advertisement(LSA)에 다음과 같은 추가 정보가 필요하다. max. link bandwidth max. reservable link bandwidth current bandwidth reservation current bandwidth usage link coloring 등

트래픽 할당 Mapping traffic to paths load balancing의 과제 congestion 해소 loss/delay 최소화 traffic partitioning 혹은 traffic aggregation 1-traffic stream to n-paths n-traffic streams to 1-path n-traffic streams to m-paths

경로 관리 Traffic monitoring과 measurement path 상의 고장(failure)의 발견과 이에 대한 대응 대체 LSP의 선택, rerouting 즉, 망 보호/복구(protection/restoration) 기능

목차 IP Traffic Engineering IP QoS Traffic Engineering 구성 요소 경로 설정과 Constraint-based routing TE signaling과 MPLS IP QoS Motivation Approach Intserv RSVP Diffserv Reference

Motivation Characteristics of IP New Internet Applications Connectionless Best-Effort Service = No QoS New Internet Applications Multimedia Applications with various QoS Requirements Bandwidth, Delay, Loss, …

Approach QoS Service Model QoS Signaling Protocol Others Extend the Traditional Best-Effort Service Model Intserv (Integrated Services), Diffserv (Differentiated Services) QoS Signaling Protocol Reserve Resources for QoS-Requiring Flows RSVP (Resource ReSerVation Protocol) Others Admission Control, Queuing Discipline, ...

Intserv Characteristics QoS Service Type Per-Flow based End-to-End QoS Guarantee Need Per-Hop Signaling Protocol Not Scalable QoS Service Type Guaranteed Service Loss-intolerant and hard real-time service Guaranteed delay bound with no queuing loss Controlled-Load Service Best-effort service under unloaded condition No specific delay & loss guarantee

RSVP 1/2 QoS Signaling Protocol Heterogeneous Receiver-Controlled Simplex Multipoint Reservation Request by Flow Descriptor = (Flowspec, Filterspec) Soft State in routers to adapt to routing changes Need Periodic Refresh  Not Scalable  RSVP Optimization Flexible control over sharing of reservations and forwarding of subflows PATH, RESV Messages

RSVP 2/2 Optimization Bundle Message MESSAGE_ID Extension Aggregate multiple RSVP messages in one PDU Reduce overall message handling overhead MESSAGE_ID Extension Shorthand indication of a refresh message  Allow receiver to readily identify an unchanged message Reduce refresh message processing overhead Summary Refresh Message Send Srefresh message with a list of Msg_ID fields of corresponding PATH/RESV trigger messages Do not need to transmit whole refresh messages

Diffserv 1/9 Characteristics Aggregate QoS Service No Per-Flow State, No Per-Hop Signaling DSCP Marking & Traffic Conditioning at Edge Scalable Service Discrimination Aggregate QoS Service DS Field DSCP (DS CodePoint; 6 bits) CU (Currently Unused; 2 bits) IPv4 TOS Field, IPv6 Traffic Class Field DSCP Mapped to PHB (Per-Hop Behavior) PHB : Externally observable forwarding behavior applied at a DS-compliant node to a DS BA

Diffserv 2/9 Aggregation of flows Classification and conditioning LSR LSR LSR LSR Many flows LSR LSR Aggregation of flows LSR Classification and conditioning (metering, marking, policing 등) Per Hop Behavior (scheduling, dropping) codepoint PHB class

16-bit total length (in byte) Diffserv 3/9 IPv4 Header 4-bit header length 8-bit type of service (TOS) 4-bit version 16-bit total length (in byte) Currently Unused 6-bit DS codepoints 2-bit CU PHB codepoints Default 000000 class selector xxx000 EF 101110 AF

Diffserv 4/9 PHB (Per Hop Behavior) Default PHB Class Selector PHB Best-Effort Forwarding Class Selector PHB Relative Order, Compatible with IP Precedence Field EF (Expedited Forwarding) PHB (Configured) Min Departure Rate > Max Arrival Rate  Virtually No Queueing Used to Build a Low Loss, Low Latency, Low Jitter, Assured Bandwidth, End-to-End Service (Virtual Leased Line Service) AF (Assured Forwarding) PHB Group N AF Classes, M Drop Precedence Levels/Class No Quantifiable Timing Requirement

Diffserv 5/9 Traffic Conditioning Control Functions Applied to a Behavior Aggregate To Enforce Agreements between Domains and To Condition Traffic to Receive a Differenciated Service within a Domain By Marking packets with the appropriate Codepoint and By Monitoring and Altering the Temporal Characteristics of the Aggregate Meter  Classifier Marker  Shaper/ Dropper  : Components of Traffic Conditioner

Diffserv 6/9 PDB (Per Domain Behavior) Expected Treatment that an Identifiable or Target Group of Packets will receive from “Edge to Edge” of a DS Domain Associated with PHB (or List of PHBs) Traffic Conditioning Requirements PDB Type VW (Virtual Wire) PDB BH (Bulk Handling) PDB

Diffserv 7/9 VW (Virtual Wire) PDB Edge-to-Edge Service  Dedicated Wire The aggregate input rates are appropriately policed and the EF service rates on interior links are appropriately configured At every transit node, the aggregate’s maximum arrival rate is less than that aggregate’s minimum departure rate Premium Service, or Virtual Leased Line (VLL) EF PHB is used Concerns : Routing stability, Multipath routing  5~10% fudge factor in reality

Diffserv 8/9 VW PDB (cont’d) Parameterization  Traffic : packet size, physical wire bandwidth Domain : link bandwidth, EF transit bound of a router  Maximum VW bandwidth allocated between ingress & egress routers  packet size / EF bound When the ingress VW flows are aggregated/policed appropriately, the EF BA can consist of an arbitrary aggregate of VW flows

Diffserv 9/9 BH (Bulk Handling) PDB For Extremely Non-Critical Traffic Packets of BH PDB may be delayed or dropped when any other traffic is present Packets of BH PDB are forwarded onto the node output link when the link is idle Intended as an Additional Tool for Administrators in Engineering Networks No Attributes, No Parameters Class Selector (CS1) PHB, AF PHB, or EXP/LU PHB can be used

Reference Intserv RSVP Diffserv RFC 1633, 2210, 2211, 2212 RFC 2205, draft-ietf-rsvp-reduct-05 Diffserv RFC 2474, 2475, 2597, 2598 draft-ietf-diffserv-pdb-def-03, draft-ietf-diffserv-pdb-vw-00, draft-ietf-diffserv-pdb-bh-02