for guaranteed IP datagram routing



From this document you will learn the answers to the following questions:

What is the method that only border routers are aware of the bandwidth management?

What does the distributed bandwidth management control of IP datagram networks have?

The core router needs to maintain the state information on the LSP including the incoming and what labels?

Similar documents
Investigation and Comparison of MPLS QoS Solution and Differentiated Services QoS Solutions

Can Forwarding Loops Appear when Activating ibgp Multipath Load Sharing?

How To Make A Network Plan Based On Bg, Qos, And Autonomous System (As)

A Review on Quality of Service Architectures for Internet Network Service Provider (INSP)

Exterior Gateway Protocols (BGP)

Quality of Service Routing in MPLS Networks Using Delay and Bandwidth Constraints

Internet Firewall CSIS Packet Filtering. Internet Firewall. Examples. Spring 2011 CSIS net15 1. Routers can implement packet filtering

How To Provide Qos Based Routing In The Internet

MPLS/BGP Network Simulation Techniques for Business Enterprise Networks

Introducing Basic MPLS Concepts

Supporting End-to-End QoS in DiffServ/MPLS Networks

Implement a QoS Algorithm for Real-Time Applications in the DiffServ-aware MPLS Network

Integrated Service (IntServ) versus Differentiated Service (Diffserv)

Routing with OSPF. Introduction

Module 7. Routing and Congestion Control. Version 2 CSE IIT, Kharagpur

Enterprise Network Simulation Using MPLS- BGP

Multi-Protocol Label Switching To Support Quality of Service Needs

Building MPLS VPNs with QoS Routing Capability i

An Analysis of the DiffServ Approach in Mobile Environments

Border Gateway Protocol BGP4 (2)

BGP Best Path Selection Algorithm

Border Gateway Protocol (BGP)

Class of Service (CoS) in a global NGN

QoS Performance Evaluation in BGP/MPLS VPN

CLASSLESS INTER DOMAIN ROUTING - CIDR

Inter-domain Routing Basics. Border Gateway Protocol. Inter-domain Routing Basics. Inter-domain Routing Basics. Exterior routing protocols created to:

Internet Quality of Service

Inter Domain Routing Working Group Chemnitz University of Technology Intended status: Standards Track July 7, 2008 Expires: January 8, 2009

MPLS. Packet switching vs. circuit switching Virtual circuits

A Preferred Service Architecture for Payload Data Flows. Ray Gilstrap, Thom Stone, Ken Freeman

Faculty of Engineering Computer Engineering Department Islamic University of Gaza Network Chapter# 19 INTERNETWORK OPERATION

Introduction to TCP/IP

Route Discovery Protocols

Active measurements: networks. Prof. Anja Feldmann, Ph.D. Dr. Nikolaos Chatzis Georgios Smaragdakis, Ph.D.

MPLS Architecture for evaluating end-to-end delivery

PRASAD ATHUKURI Sreekavitha engineering info technology,kammam

OSPF Version 2 (RFC 2328) Describes Autonomous Systems (AS) topology. Propagated by flooding: Link State Advertisements (LSAs).

basic BGP in Huawei CLI

Inter-domain Routing. Outline. Border Gateway Protocol

Increasing Path Diversity using Route Reflector

4 Internet QoS Management

Outline. EE 122: Interdomain Routing Protocol (BGP) BGP Routing. Internet is more complicated... Ion Stoica TAs: Junda Liu, DK Moon, David Zats

Interdomain Routing. Project Report

18: Enhanced Quality of Service

Quality of Service for VoIP

Using the Border Gateway Protocol for Interdomain Routing

An Effective approach to control Inter-domain Traffic Engineering among Heterogeneous Networks

Routing Protocols OSPF CHAPTER. The following topics describe supported routing protocols. Topics include OSPF, page 9-1 IS-IS Protocol, page 9-3

Quality of Service Mechanisms and Challenges for IP Networks

Experiences with Class of Service (CoS) Translations in IP/MPLS Networks

Constructing End-to-End Traffic Flows for Managing Differentiated Services Networks

20. Switched Local Area Networks

6.263 Data Communication Networks

A Prototype Implementation of the Two-Tier Architecture for Differentiated Services

Using OSPF in an MPLS VPN Environment

How To Share Bandwidth On A Diffserv Network

IP Traffic Engineering over OMP technique

CS/ECE 438: Communication Networks. Internet QoS. Syed Faisal Hasan, PhD (Research Scholar Information Trust Institute) Visiting Lecturer ECE

Based on Computer Networking, 4 th Edition by Kurose and Ross

QoS Strategy in DiffServ aware MPLS environment

Research and Development of IP and Optical Networking

The Internet. Internet Technologies and Applications

Routing Protocols. Interconnected ASes. Hierarchical Routing. Hierarchical Routing

Routing architecture in DiffServ MPLS networks

Project Report on Traffic Engineering and QoS with MPLS and its applications

Dynamic Routing Protocols II OSPF. Distance Vector vs. Link State Routing

DMT: A new Approach of DiffServ QoS Methodology

Differentiated Services:

DEMYSTIFYING ROUTING SERVICES IN SOFTWAREDEFINED NETWORKING

Internet Control Message Protocol (ICMP)

A Fuzzy Algorithm for QoS-Based Routing in MPLS Network

WAN Topologies MPLS. 2006, Cisco Systems, Inc. All rights reserved. Presentation_ID.scr Cisco Systems, Inc. All rights reserved.

Quality of Service for IP Videoconferencing Engineering White Paper

IP interconnect interface for SIP/SIP-I

MPLS VPN Services. PW, VPLS and BGP MPLS/IP VPNs

Network Level Multihoming and BGP Challenges

MPLS Basics. For details about MPLS architecture, refer to RFC 3031 Multiprotocol Label Switching Architecture.

A Fast Path Recovery Mechanism for MPLS Networks

Router and Routing Basics

MPLS Multiprotocol Label Switching

Opnet Based simulation for route redistribution in EIGRP, BGP and OSPF network protocols

Addition of QoS Services to an MPLS-enabled Network

Datagram-based network layer: forwarding; routing. Additional function of VCbased network layer: call setup.

Multiple Layer Traffic Engineering in NTT Network Service

Multiprotocol Label Switching (MPLS)

QoS Implementation For MPLS Based Wireless Networks

Lecture 8: Routing I Distance-vector Algorithms. CSE 123: Computer Networks Stefan Savage

Quality of Service in the Internet. QoS Parameters. Keeping the QoS. Traffic Shaping: Leaky Bucket Algorithm

Adaptive Measurement Based QoS Management in DiffServ Networks

Network-Wide Prediction of BGP Routes

PART III. OPS-based wide area networks

Quality of Service (QoS)) in IP networks

Figure 1: Network Topology

Lecture 18: Border Gateway Protocol"

Implementing MPLS VPN in Provider's IP Backbone Luyuan Fang AT&T

How To Understand Bg

CSC458 Lecture 6. Homework #1 Grades. Inter-domain Routing IP Addressing. Administrivia. Midterm will Cover Following Topics

Introduction to Differentiated Services (DiffServ) and HP-UX IPQoS

Internet inter-as routing: BGP

Incremental QoS Deployment based on Network Brokers

Transcription:

Core stateless distributed admission control at border routers for guaranteed IP datagram routing Takahiro Oishi Masaaki Omotani Kohei Shiomoto NTT Network Service Systems Laboratories, NTT corporation -9- Midori, Musashino, Tokyo 80-8585, Japan Phone: +8 4 59 4645/ Fax: +8 4 59 4549 E-mail:{ooishi.takahiro,Omotani.Masaaki,Shiomoto.Kohei}@lab.ntt.co.jp} Abstract This paper proposes a distributed bandwidth management control for high-speed IP datagram networks. Each border router maintains topology database including shortest path tree to other border routers and reserved bandwidth on each link. User requests traffic demand to the border router, in which it is accommodated. The border router checks whether sufficient bandwidth can be reserved along with the shortest path tree originating from the border router to all possible border routers to provide the traffic demand. If sufficient bandwidth can be reserved, the traffic demand is admitted. Otherwise it is rejected. The requested traffic demand is notified between border routers via BGP-4 so that other border routers can perform the same admission decision. Thereby the admission decision is performed at each border router in a distributed manner. The proposed method can be applied to large-scale Internet backbone network. We demonstrate the proposed bandwidth management control is simple yet efficient through numerical examples. Introduction Differentiated service (diffserv) model is discussed for scalable Internet QoS service architecture in the core network[]. In the diffserve model, treatment of IP packet at the core router is associated with the special IP header field, i.e., diffserv code point (DSCP). The association between treatment and DSCP is referred to as per-hop behavior (PHB). There are three PHB classes already defined: expedited forwarding (EF), assured forwarding (AF), and best effort(be). The EF class is designed to implement the virtual leased-line(vll) service in connection-less IP datagram forwarding networks. The sufficient bandwidth is reserved so that the rate of incoming flows should not exceed the rate of the outgoing flows at each hop in the core network minimizing the queueing delay in the core network. Service level specification (SLS) is contracted between user and network. Traffic demand can be included in the SLS. The traffic injected into the network is enforced at the ingress border router of the network (See Fig). Such network resource as bandwidth and buffer is reserved to maintain the SLS in the network. Network shaper : border router : core router Figure: Diffserv Internet QoS model In diffserv, the resource provisioning for s is performed by the bandwidth broker (BB)[6]. (See Fig. ) RARs Network BB service users SLSs Network5 BB Figure: Bandwidth Broker (BB) Network4 Network A BB is set up at each domain, and a BB manages the QoS resources within a given domain based on the SLS in each domain. So the BB gathers and monitors the state of QoS resources within its domain and on the edges of the adjacent domains. When users want to allocate request bandwidth, Resource Allocation Request(RAR) is issued, and the BB of the user s domain responsively allocates the resource based on the SLS. To provide sufficient bandwidth to maintain the traffic demand, MPLS-diffserv interwork was proposed [][]. Label switched path (LSP) is established, along with SLSs Inter-domain Communication Intra-domain Communication BB

which the sufficient bandwidth is provided. This method requires the core router to handle MPLS protocol. Signaling and forwarding need to be implemented in the core network. The core router needs to maintain the state information on the LSP including the incoming and outgoing labels and the required bandwidth. The core router should be as simple as possible to be able to be used in large-scale Internet backbone. The MPLS-diffserv interwork method has another drawback. The destination address may not be necessarily specified in the SLS. Suppose that there is a customer who needs the specific bandwidth to the commercial web server to minimize the response time. The customer may not be able to specify all destination addresses of potential users. In this situation it is very difficult to specify the reserved bandwidth in a point-to-point fashion (also known as "pipe model") rather than in a point-to-any fashion (also known as "hose model")[4]. In this paper we propose a distributed bandwidth control method to make admission decision for the new flow requesting EF class. Only border routers are aware of the bandwidth management in the propose method. Only the IP datagaram packet delivery with PHB is required in the core routers. The border router maintains the shortest-path tree to all possible other border routers and traffic demand associated with it. The border router calculates the residual bandwidth on receipt of user traffic demand. The user traffic demand is distributed among the border routers by using the internal border-gateway protocol () sessions. Thereby the proposed method can be applied for the large-scale Internet backbone. The rest of the paper is organized as follows. In Chapter we propose a distributed bandwidth control method. In Chapter we demonstrate the effect of the proposed method through quantitative analysis. In Chapter 4 we draws the conclusions and address the future research items. Distributed bandwidth management control.. SA-SPT There are three types of routers in the network: access, border, and core routers. Access router is directly connected to user facilities i.e., hosts and routers. It is directly connected to more than one border routers and is not directly connected to any core routers. Border router is located at the border of the network. Border routers are located between access router and core router. Border router can be connected to access, border, and core routers. Border router is used to connect other network: border router is connected with border router in the other network. (See Fig. ). In this paper we assume that interior gateway protocol (IGP) is used for routing in the core network. The border and core routers speak IGP. The shortest path is selected by the IGP. Exterior gateway protocol (EGP) such as BGP-4 is used to exchange the reachable network address prefix between access and border routers in the same network and between border routers of the adjacent networks. We also assume that the reachable network address prefix is propagated using internal BGP () session [5]. User's SLS includes traffic demand. Sufficient bandwidth is reserved in the network to guarantee the traffic demand. If the specific destination address is included in the user's SLS, the shortest path from the source border router to the destination border router is calculated and the sufficient bandwidth is reserved along with the path. If the specific destination address is not included in the user's SLS, the shortest path tree originating from the source border router to all possible destination border routers is calculated and the sufficient bandwidth is reserved along with the tree, which we refer to as the SA-SPT (source-border-router to all-border-router shortest path-tree). In this way we consider the worst-case scenario unless the specific destination is included in the user's SLS. Figure explains the SA-SPT. Suppose that a user connected to the border router is requesting the traffic demand x [Mb/s]. We calculate the SA-SPT, the shortest-path tree originating from the to other border routers,, and 4. We assume that the traffic demand is offered to all the links in the SA-SPT. In Figure, the SA-SPT is depicted in solid line while the physical links not used in the SA-SPT are depicted in dashed line. Each border router has the complete topology database of the network because link-state type IGP is used in the network. Each border router tells how much bandwidth need to be reserved for the users it is directly connected to via BGP-4 as mentioned later. In this way it calculates the reserved bandwidth for all request from all border routers on each link in the network in a distributed manner. By subtracting the reserved bandwidth from the link capacity, it tells how much bandwidth can be allocated for a new request.

AR [Mb/s] SA SPT from to all s physical links Figure : Shortest path tree originating from the border router to all other border routers.. Admission decision algorithm description Notations are introduced to describe the admission decision algorithm formally : 4 } Bk=0 ek E for all i s.t. vi V {vy}{ for all k s.t. ek Ei{ Bk= Bk+Fi }continue k }continue i admission decision(about vy) i = vy for all k s.t. ek Ei{ Ri= min(ck Bk) } Given parameters: If Ri 0 Ri is acceptable bandwidth from i V={vi}; the set of nodes E={ek}; the set of links G=(V,E) ; all topology Hi=(Vi,Ei) ; SA-SPT topology from i V; the set of s V; the set of s N; the number of (= V ) N; the number of (= V ) Bk; reserved bandwidth of link k Ck; link capacity of link k Fi; input traffic from i Variable parameters: vy; which acts admission decision Ri; acceptable bandwidth from i Definition of set operation V {vy}= V {vy} Admission decision algorithm is formally described in what follows. reserved bandwidth calculation for all k{.. Bandwidth information exchange between border routers Each border router needs to know traffic demand from, which all border routers are requested to provide to their users. The user requests the traffic demand only to the border router it is accommodated by. Mechanism to notify the traffic demand to other border routers is required.... Traffic demand exchange via BGP-4 BGP-4 is a EGP, which is used to notify the network address prefix between autonomous systems. Network address prefix is learned from the adjacent autonomous system via BGP-4. The learned prefix is then advertised to the other border routers in the autonomous system via sessions. BGP-4 carries attributes associated with each network prefix to perform policy routing. A set of attributes defined in the standard document includes LOCAL PREFERENCE, MED, AS_PATH, and COMMUNITY. We defined a new attribute by extending BGP-4. It is a BANDWIDTH_AGGREGATE () attribute.... attribute The attribute is used to notify the total traffic demand injected to the border router. Figure 4 shows how the attribute is used. The total bandwidth injected

attribute from the is denoted by and is notified as the attribute. In this way the total traffic demand injected from is notified to all the other border routers,,, and 4. The same mechanism is used for traffic demand injected from,, and 4. By using this mechanism, the complete information on the traffic demand from all border routers is shared by all border routers. Thereby the border router can calculate the residual bandwidth in a distributed manner. 9.60.. AS EBGP AS [Mb/s] [Mb/s] 9.60.. B. Performance evaluation attribute Figure4: bw_agr attribute Bandwidth efficiency achieved by the proposed method is examined. We applied the proposed method to the mesh network shown in Fig. 5. We calculated the admissible traffic demand from 5 under the assumption that the traffic demand from border routers except 5 is identical x [Mb/s]. The shortest path, which is calculated from the cost of each link shown in Fig. 6, is used for packet routing. We assume that the capacity of each link is 50 [Mb/s]. The relationship between the admissible traffic load from 5 y and ones from all the other border routers x is shown in Fig. 6. We observe that the admissible traffic from 5 is decreasing at the small rate as the traffic demand from the other border routers increase. attribute 9.60.. AS5 9.60.. 4 AS [Mb/s] 4 [Mb/s] AS4 Figure 5 : Network model Figure 6: Bandwidth efficiency achieved by the proposed method.. Link failure 7 9 acceptable bandwidth (y[mb/s]) 50 40 0 0 0 8 0 In Fig. 7, we show the admissible traffic load from 5 by a dotted line when one link failure occurs in the topology which shows with in Fig. 5. The admissible traffic load decreased in comparison with the case of no link failure. As a result of link failure from to 7, this admissible traffic load decrease was lead. This analysis is future work. 6 5 4 y [Mb/s] = α acceptable bandwidth from 5 y = x 0 0 5 0 5 0 traffic demand ( [Mb/s]) α

acceptable bandwidth (y[mb/s]) 50 acceptable bandwidth from 5 [6] L. Dunn, R. Neilson, V. Narayan, F. Reichmeyer, B. Teitelbaum, S. Hares, Internet QBone:building a testbed for differentiated services, IEEE Network, Sep./Oct. 999. 40 0 0 α 0 y = x 0 0 5 0 5 0 traffic demand ( [Mb/s] ) Figure7: Bandwidth efficiency achieved by the proposed 4. Conclusion method when a link failure occurs We proposed the distributed bandwidth management method for datagram network. The total traffic demands injected from all border routers are notified among border routers each other via BGP-4. They are maintained and used to make admission decision for the traffic demand by each border router in a distributed way. The proposed method does not require any new mechanism in core routers. Thereby the backbone network element can be simplified and intelligent mechanism is required only at border routers. We argue that the proposed method is suitable for future high-speed IP datagram backbone network architecture. References [] S. Blake, D. Black, M.Carlson, E. Davies, Z. Wang, and W. Weiss, An architecture for differentiated services, RFC475, Dec.998. [] P. Ford, F. Baker, Y. Bernet, R. Yavatkar and L.Zhang, A framework for end-to-end QoS combining RSVP/IntServ and differentiated services, draft-bernet-intdiff-00.txt, Mar.998. [] G. Huston, Internet performance survival guide, John Wiley & Sons, Inc., edition, Feb. 000. [4] B.S. Davie and Y. Rekhter, MPLS technology and applications, Morgan Kaufmann Publishers, edition, May 000. [5] H. Ballabi and D. McPherson, Internet routing architectures, Cisco Press, edition, Jan. 000.