CHAPTER 1 INTRODUCTION

Transcription

1 CHAPTER 1 INTRODUCTION 1.1 TRAFFIC MANAGEMENT Internet is the most powerful tool of the modern world. Internet helps the world to communicate, share and explore the knowledge. In Internet, there are lot of communications in the form of mail, chat and browsing. Lot of requests and responses are being exchanged in Internet in the form of control overhead and transmission data. This data (either control overhead or transmission data) can produce a huge number of packets in Internet. Such large data may cause over load which leads to slowing down the operations of the communication devices, causing failure to respond for requests, or drop the packets. A packet-switched network is considered as a network of queues. The transmitting nodes are constantly adding packets and the receiving nodes are removing them from the queue. In a situation when, too many packets are present in the queue that is, transmitting nodes constantly pouring packets at a higher rate than the receiving nodes removing them from the queue will degrades the performance and such situation is termed as congestion. Congestion can occur due to several reasons. For example, if stream of packets arrive on several input lines and need to be out on single output line, then long queue will be build up for that line. If there is insufficient memory to hold these packets, then the packet will be dropped. Slow processors also cause congestion. If the router CPU is slow in performing the task required for them, queue is build up even if there is an excess of line capacity. Similarly, low bandwidth line can also cause congestion. Upgrading the line but not changing the processor or vice-versa will shift the bottle neck to some other point.

2 Congestion in the Internet is dynamic in nature. For solving such dynamic problems, static solutions are insufficient. Hence Traffic management becomes a most important research activity since last few decades. The traffic management lies in the layer 3 (Network Layer) of the Open System Interconnection (OSI) Model. The traffic management has three important tasks namely 1) Congestion Avoidance, 2) Congestion Control and 3) Flow Control. The congestion control and avoidance are implemented in both Transmission Control Protocol (TCP) and User Datagram Protocol (UDP). Whereas, the Flow control is implemented only in TCP as it requires the acknowledgement of the transmission data. Basically, the objective and deployment of congestion control and avoidance algorithms are almost same. The TCP congestion avoidance [32] algorithm is the basis for congestion control in the Internet. Transmission Control Protocol (TCP) uses a network congestion avoidance algorithm that includes various aspects of an Additive Increase/Multiplicative Decrease (AIMD) scheme with other schemes such as slow-start. To avoid congestion collapse, TCP uses a multi-faceted congestion control strategy. For each connection, TCP maintains a congestion window, limiting the total number of unacknowledged packets that may be in end-to-end transit. TCP uses a mechanism called slow start to increase the congestion window after a connection is initialized and after a timeout. It starts with a window of one or two segment and increases it by one segment size for each acknowledgement received. Although the initial rate is low, the rate of increase is very rapid: for every packet acknowledged, the congestion window increases by 1 MSS so that the congestion window effectively doubled for every round trip time (RTT). When the congestion window exceeds a Slow Start Threshold (SSThresh), the algorithm enters the congestion avoidance state. In some implementations (e.g., Linux), the initial SSThresh is large and so the first slow-start usually ends after a loss. However, SSThresh is updated at the end of each slow-start and often affect the subsequent slow starts triggered by timeouts. 2

3 For effective congestion control and avoidance, traffic control measure is the basic requirement. There are two types of Traffic control measures [33], 1) Macroscopic level and 2) Microscopic level. These types of traffic control measures emphasize on single point traffic control rather considering the whole or entire network. Macroscopic level of traffic control is accomplished by restricting route topology or connectivity information by the use of Border Gateway Protocol (BGP). It sometime uses Interior Gateway Protocol (IGP). The implementation in BGP has per peer announced and it accepts filters. Filters consider the route prefix. The microscopic level controls traffic at the point of forwarding instead of information propagation point. It examines packet headers and sometimes payload. It is used in firewalls and Intrusion Detection System with active response mechanisms. 1.2 TRAFFIC MEASUREMENT Traffic measurement is a pre-process stage for implementing an effective traffic management. The basic thing to motivate the traffic measurement [25] [152] is the requirement by service providers and service users. Different tasks like monitoring, debugging, engineering and architecture development requires efficient use of network that leads to the measurement of traffic. There are many reasons to measure network traffic which includes Identifying the traffic patterns Finding the traffic intensity Deciding when more capacity is needed An improved understanding of what is happening inside the network and allow improving the network performance. Most network hardware vendors are providing some network management capability in their equipment. For example, they may support the Simple Network Management Protocol and implement some of the IETF standard Management 3

4 Information Bases (MIBs). These can be used to monitor the amount of data flowing across the network links. But it is not enough to give a detailed view of the traffic and the way it behaves. The specialized methods or tools are required to get more detailed information about the traffic. These methods allow measuring things like Link behaviour (packet loss, transit time and jitter), Routing topology (routes available, routing stability) and Traffic flows (how many packets or bytes flow). The basic performance metrics of traffic includes Packet loss Delay Throughput Availability Other factors that deals with the requirement of traffic measurement are Pricing Service level agreements New services Applications The traffic can be measured in two ways. One is Active measurement and another one is Passive measurement Active Measurement In this measurement, users or providers are directly related to the activities of the measurement. There are a number of ways to carry out measurement using 1. Injection of probes into network by users and providers 2. Ping and Trace route Path connectivity Round-trip delay 4

5 3. User-application performance as seen from hosts Loss Delay Throughput Passive Measurement In this approach user indirectly deals with a system using some hardware or software tools. Basically some historical data is used to find the current traffic measurement. The intelligent devices installed in network analyse router or switch, which can provide network internal behaviour. The techniques currently used for this type of measurement are as follows 1. Packet monitors This can be achieved by recording packet headers on link It requires unique detail of protocol and architecture studies 2. Router / Switch traffic statistics Packet drops Counts Flow statistics 3. Server and router logs These logs provide summaries of dial session, routing updates, etc. The usage of Internet is increasing day by day which affects the Quality of Service (QoS). The capability to provide resource assurance and service differentiation in a network is often referred to as a Quality of Service (QoS) [160]. ISP has to provide good quality of service, must bring operational efficiencies and differentiate their service offerings to their customer within their infrastructure in a cost effective manner. 5

6 The following traffic characteristcs [121] are having impact on the QoS guarantees. 1. Traffic Variations 2. Flow and Sessions 1.3 OVERVIEW OF TRAFFIC ENGINEERING ISPs are rapidly deploying more network infrastructure and resources to handle the emerging applications and growing number of users enhancing the performance of an operational network at both the traffic and the resource levels [19]. Traffic Engineering (TE) broadly relates to the performance optimization of an operational IP network. A major objective of traffic engineering is to minimize or eliminate high-loss situations. Another goal of traffic engineering is to balance the Quality of Service (QoS) against the cost of operating and maintaining the network. The goal of performance optimization of operational networks [18] is accomplished by routing traffic by utilizing the network resources efficiently and reliably. A routing specifies how to route the traffic between each origin-destination pair across a network. The traffic sharing is applied in routing and allocating process to enhance the survivability of network [157]. There are two approaches in traffic engineering. One is Overlay approach and the other is Integrated approach. The overlay approach [60] is used by many service providers for traffic engineering. In this approach, the service providers establish logical connections between the edge nodes of backbones and then mapping these logical connections onto the physical topology and form a full mesh virtual network overlaying on top of the physical topology. 6

7 Service providers can control the distribution of traffic over physical topology by carefully routing these logical connections over physical links. The optimal mapping between the logical connections and the physical links can be computed using a linear programming formulation. Wang et al [151] proposed an approach called integrated approach. The integrated approach accomplishes traffic engineering objectives without full mesh overlaying. Instead of overlaying IP routing over the logical virtual network, the integrated approach runs shortest-path IP routing natively over the physical topology. For any given traffic demand, it is possible to select a set of link weights such that the shortest paths based on the selected link weights produce the same traffic distribution as that of the overlay approach with the assumption that traffic between the same source destination pair can be split across multiple equal cost shortest paths, if they exist. To meet the traffic engineering objectives, it is needed to place the demands over the links in such a way that the traffic distribution is balanced and there is no congestion or hot spot in the network. 1.4 MULTI-PROTOCOL LABEL SWITCHING (MPLS) NETWORK MPLS is a mechanism used in high-performance telecommunication networks that directs and carries data from one network node to the next with the help of labels. MPLS is an additional effort and functionalities which is added to improve the conventional IP routing architecture and protocols such as new connectivity abstraction. In point-to-point paths, Label Switched Paths (LSP), can be established using label based forwarding mechanisms. MPLS operates at an OSI Model layer that is generally considered to lie between traditional definitions of Layer 2 and Layer 3 and thus is often referred to as a Layer 2.5 protocol, which is shown in the Figure 1.1 below. 7

8 Application L Presentation L Session Layer Transport Layer Network Layer Data Link Layer MPLS Operations Physical Layer Figure 1.1 Operation of MPLS in OSI layer The MPLS network was designed to provide an unified data-carrying service for both circuit-based clients and packet-switching clients which gives a datagram service model. The MPLS network can be used to carry different kinds of traffics, including IP packets, as well as native ATM, SONET and Ethernet frames. MPLS is currently replacing some of these technologies in the marketplace. It is highly possible that MPLS will completely replace these technologies in the future, thus aligning these technologies with current and future technology needs. MPLS was originally proposed by a group of engineers from IPSILON Networks, but their IP Switching technology, which was defined only to work over ATM, did not achieve market dominance. Cisco Systems, Inc., introduced a related proposal, not restricted to ATM transmission, called Tag Switching. It was a Cisco proprietary proposal and was renamed Label Switching. It was handed over to the Internet Engineering Task Force (IETF) for open standardization. The IETF work involved proposals from other vendors and development of a consensus protocol that combined features from several vendors' works. 8

9 This ensures end-to-end circuits over any type of transport medium using any network layer protocol. In view of the fact that MPLS supports Internet Protocol [28] revised versions (IPv4 and IPv6), IPX, AppleTalk at Layer3, Ethernet, Token Ring, Fiber Distributed Data Interface (FDDI), Asynchronous Transfer Mode (ATM), Frame Relay and PPP (Point to Point Protocol) at Layer 2 and WDM Ring Networks [48]. The recent proposals like embedded MPLS [72], Green MPLS [79] are recommended for further studies. For functionality, management and improved MPLS architecture [124] is recommended. 1.5 DATA STRUCTURE AND PACKET FORMAT OF MPLS NETWORK MPLS works by prefixing packets with an MPLS header, containing one or more labels. This is called a label stack. Each label stack entry contains four fields: A 20-bit label value A 3-bit Traffic Class field for QoS (quality of service) priority (experimental) and ECN (Explicit Congestion Notification) A 1-bit flag at the bottom of stack. If this is set, it signifies that the current label is the last in the stack An 8-bit TTL (time to live) field These MPLS-labelled packets are switched after a label lookup/switch instead of a lookup into the IP table. As mentioned above, when MPLS was conceived [37], label lookup and label switching were faster than a routing table or Routing Information Base (RIB) lookup because they could take place directly within the switched fabric and not the CPU. The entry and exit points of an MPLS network are called Label Edge Routers (LER), which push an MPLS label onto an incoming packet and pop it off the outgoing packet respectively. Routers that perform routing based only on the label are called Label Switch Routers (LSR). In some applications, the packet presented to the LER may have a label already, so that the new LER pushes a second label onto the packet. 9

10 1.6 MPLS ARCHITECTURE The sample MPLS architecture is shown in Figure 1.2 below. It contains Label Edge Router (LER) and Label Switch Router (LSR). The packet traverse from the LAN 1 ( /24) R1 R2 (MPLS Ingress LER)-R3 (LSR) - R4 (LSR) - R5 (Egress LER) R6 LAN 2 ( /24). Here R1- R2- R3- R4- R5- R6 refers to routers used for packet switching operation. The Elements of Layered switch are: Figure 1.2 Sample Architecture of MPLS Network Forward Equivalence Class (FEC): This class includes a cluster of packets of a specific application forwarded in its switch path over the same pathway (with same forwarding treatment). Every packet of a particular class holds the same service requirement [50]. Every type of data traffic is assigned with a new FEC and is done only once while they enter the MPLS cloud. 10

11 Ingress Label Edge Router: It exists on the perimeter of an MPLS cloud and is an entry point where the data packet originates from its source. This edge router imposes label (PUSH) and forward packets to destination through the domain. After setting up LSP and assigning labels, this ingress edge router initiates packet-forwarding process in MPLS core network. Egress Label Edge Router: It exists on the perimeter of an MPLS network and is an exit point where the data packet reaches its destination. This edge router performs label disposition or removal (POP) and forward IP packet to destination. It disposes label from the arrived packet only when bottom-of-stack indicator identifies whether the encountered label is bottom label of the stack or not. Label Switch Router (LSR): This router receives a labelled packet, swaps it with an outgoing one, and forwards the new packet to an appropriate interface. Depending on its location in MPLS domain, this router performs label disposition (removal, POP), label imposition (addition, PUSH) or label swapping (replacing the top label in a stack with a new outgoing label value). When the data stream (files or multimedia traffic) arrives from the access network to the MPLS core, it is segregated into separate FEC in this router. As an acknowledgement of label bindings, LSR creates entries in Label Information Base. This table comprises of I/O ports and I/O port labels indicating the label-fec mapping. Label Switch Path (LSP): LSP is the path traversed by a packet from source to destination through an MPLS-enabled network. The path is simplex type or one-way characteristic. This allows packets to be switched from one edge to the other by traversing several intermediate switch routers. Every network location needs LSPs to be established for data transfer. For example, packet data from LER1 traverses among several intermediate nodes to LER2 using LSP1, then another path LSP2 is set out for packet transfer to the other end directly, which is the shortest path to arrive the destination. However, path switching is derived from IGP routing information and may diverge from Interior Gateway Protocol preferred path to the target network. 11

12 1.7 MPLS HEADER fields: A 32-bit MPLS header is shown in figure 1.3 below, which contains the following Label (20bits) CoS (3bits) S (1bit) TTL (8bits) Figure 1.3 MPLS Header Format Label (20 bits): It is a short, fixed length field with a physically continuous identifier, used to classify a FEC of local significance. After this classification, the packet is given a label value, which is used for packet forwarding. These labels are created based on network topology, request and incoming traffic. Since they are associated with FEC, they undergo binding. Both the process of binding and label distribution are done by downstream routers. The labels are removed at egress and the packet is sent to destination. Class of Service (CoS) or EXP (Experimental) field: This 3-bit field influences queuing and discard algorithms applied to the packet when transmitted through the network. Since the CoS field has 3 bits, there are eight (2 3 ) distinct service classes in use. Stack Field (S): A label stack is an ordered set of labels where each operates with a specific function. If an ingress router imposes more than one label in a single IP packet then it forms a stack of labels. Here, the bottom-of-stack indicator (S) is created to mention the end of label stack. As mentioned earlier, LSR swaps only the top label in a label stack. An egress LSR continues label disposition from the stack until it finds bottom-of-stack-indicator (S). After this, a route lookup takes place using IP Layer 3 header details and the switch router forwards packet towards the destination. 12

13 Time To Live (TTL): A TTL 8-bit field performs the same function of IP s TTL, where the packet is discarded when TTL is 0 to prevent looping in the network. Whenever a labelled packet traverses through LSR, the TTL value is decremented by one. Upstream and Downstream: The concept of upstream and downstream is pivotal in understanding label distribution (control plane) and packet forwarding in MPLS domain. Both upstream and downstream are references to network. Data is downstream for a target network as it is the destination point. Updates pertaining to a specific prefix are always propagating upstream (routing protocol or label distribution). Label Distribution Protocol (LDP): This protocol distributes [8] labels in a MPLS network based on the routing information from IGP. The router s Forwarding Information Base (FIB) determines a hop-by-hop path throughout the network. Unlike traffic-engineered paths that use constraints and explicit routes to set up end-to-end Label Switched Paths, LDP is used only for signalling the best effort LSPs. LDP discovery enables a Label Switch Router to determine the potential of its peers. Basic mechanisms such as discovery and extended discovery are used to locate LSRs at link level. 1.8 TRAFFIC ENGINEERING IN MPLS NETWORK Traffic engineering is the major research area in MPLS due to the emerging requirements of MPLS and the internet usage. Traffic engineering is a method of optimizing the performance of a telecommunications network by dynamically analyzing, predicting and regulating the behaviour of data transmitted over that network. Traffic engineering is also known as tele-traffic engineering and traffic management. The techniques of traffic engineering [39] can be applied to networks of all kinds, including the Public Switched Telephone Network (PSTN), Local Area Networks (LANs), Wide Area Networks (WANs), cellular telephone networks, proprietary business and the Internet. Traffic at the packet level for different applications tends to have different characteristics. This fact has been observed for emerging applications such as video 13

14 conferencing, peer-to-peer and multimedia applications. Moreover, heterogeneous traffic streams are multiplexed together to share the same link. When traffic of different characteristics is multiplexed together, traffic distortion occurs which can be significant depending on the characteristics of the streams MPLS network with label-switched paths, potentially for different heterogeneous streams. If multiple paths are being aggregated to share bandwidth, then all of such paths together would be considered as a tunnel. The tunnel is the unit of flow to which a certain bandwidth is allocated at each router that it traverses. In order to minimize the distortion, a possible approach would be applied to classify and multiplex appropriate streams into tunnels. That is, like-minded streams can avoid (or minimize) distortion if the network has the capability to do so. For this purpose, one can consider the MPLS network in which multiple LSP tunnels can be set up between source and destination nodes; such tunnels can be used to ensure logical separation between streams in order to minimize distortions [52]. When message queues become unacceptably long or the frequency of busy signals becomes unacceptably high, the network is said to be in a high-loss condition [56]. The traffic engineering objective is to minimize the high-loss situations and to balance the Quality of Service (QoS) against the cost of network [17]. The QoS requirement of a connection can be given as a set of link constraints, for example by requiring that there is enough bandwidth on the path selected for the user requirements. The capacity needed at MPLS layer is provided by the underlying transport network which may be based on Synchronous Digital Hierarchy (SDH) or Wavelength Division Multiplexing (WDM), depending on the link speed. The transport network can be devoted to MPLS services only, or, more often, shared with circuit switched services such as the phone service. A node supporting MPLS is named LSR. Edge nodes must necessarily support MPLS in order to collect packet switched traffic from users. Core nodes may or may not support MPLS. LSRs and virtual links connecting them define a logical network topology on top of the physical topology of the transport network. Virtual links in the logical topology are mapped into paths of physical links in the physical topology. These paths between LSRs are circuits 14

15 (or light-paths) and may cross several nodes of the transport network not supporting MPLS. Circuits must be dimensioned according to bandwidth requirements and then the capacity of each physical link must be selected based on the circuits crossing it and the discrete set of possible values defined by the transport technology. From the network technology perspective, the integration of the optical layers with electronic layers within a converged data-optical infrastructure based on classical IP or modern GMPLS (Generalized MPLS) architecture is a key element in the current trend in broadband network evolution. Two-layer network design problems [143], where link and node dimensioning are also included in the model, have been considered only quite recently. Some works specifically consider MPLS technology and some of them address the problem of MPLS node location. Given the complexity of the optimization models, several authors rely on path formulations and column generation coupled with branch and bound, joint column and row generation methods, branch and cut with cut-set inequalities or LP-based decomposition approaches. For mid-to-large networks, the solution of choice remains heuristic algorithms, which provide a feasible solution in limited time. However, to the best of various researches, the effect of statistical multiplexing has not been previously considered in such network design and routing models. 1.9 OBJECTIVE In this thesis, a Hybrid Traffic Management (HTM) model is proposed, which contains two traffic models namely traffic flow analysis model and ant colony optimization based routing model. The traffic flow analysis model will analyse the traffic pattern and identify the low load path. The ACO is used for effective transmission in the low load path. The proposed HTM provides less packet loss and less response time than existing routing protocol. This hybrid model enables traffic free environment in the congested MPLS network. The routing packet size of the proposed HTM is higher than ACO but comparing existing traditional routing protocol like OSPF and RIP, the packet size is very less. 15

16 1.10 TOOLS USED - NETWORK SIMULATOR (NS2) For the implementation purpose, the Network Simulator (NS2) tool is used. NS2, also written as "ns2" (Network Simulator in its 2nd edition) began as a variant of the REAL network simulator in 1989 and has evolved substantially over the past few years. In 1995 NS development was supported by DARPA through the VINT project (Virtual InterNetwork Testbed) at LBL, Xerox PARC, University of California in Berkeley, University of California in San Diego and its Information Sciences Institute. NS2 is a discrete event simulator. NS2 provides substantial support for simulation of TCP, UDP, routing and multicast protocols over wired and wireless networks. NS-2 is a public domain simulator boasting a rich set of Internet Protocols, including terrestrial, wireless and satellite networks [113]. In NS-2, the following agent creation is possible Wired network Terrestrial network, Satellite network Wireless networks Various routing algorithms like DV, LS, PIM-DM, PIM-SM, AODV, DSR, DSDV. Traffic sources like web, ftp, telnet, cbr, stochastic traffic. Failures, including deterministic, probabilistic loss, link failure, etc. Various queuing disciplines (drop-tail, RED, FQ, SFQ, DRR, etc.) QoS (e.g., IntServ and Diffserv). In NS-2, the following can be visualized Packet flow, queue build up 16

17 Packet drops. Protocol behaviour such as TCP slow start, self-clocking, congestion control, fast retransmit and recovery. Node movement in wireless networks. Annotations to highlight important events. work. These features and flexibility of NS2 are the major reasons to choose NS2 for this NS2 Input All the necessary information are written in OTcl script and given as input to NS2. The Object like nodes, links, etc. is instantiated with the script. The input script defines the topology, builds the agents (sources and destinations), sets the trace files and sets the start time in the simulation NS2 Output The output is recorded by using trace or monitor objects. Trace objects collect the data for each packet. Monitor objects collect data on an aggregate level and are implemented as counters of specific parameters of interest like total number of packet or byte arrivals. To have a complete understanding of each metric pattern, tracing is performed on a per packet basis. An output trace in NS2 has a fixed format which is shown in Figure 1.4 below. Event Time From node To node Pkt type Pkt size Flags F id Source add Des. add Seq. no Pkt. id Figure 1.4 NS2 Output Trace Format 17

18 Each trace line starts with an Event (+, -, d, r) descriptor followed by the simulation time of that event. From node and to node, identifies the link on which the event occurred. The next information is packet type followed by the Flag field and Flow-id. The next two fields give the source and destination addresses. Next field contains the sequence number of the packet. The last field is the unique id of the packet. r cbr cbr d cbr r cbr cbr cbr cbr cbr r cbr cbr d cbr r cbr cbr cbr cbr r cbr cbr d cbr cbr Figure 1.5 Sample NS2 Output Trace A sample output trace is shown in Figure 1.5 above. In the above figure r represents a packet received, + represents enqueue that is Packet is added to the queue. represents dequeue that is Packet is removed from the queue and d represents a packet drop at the queue. 18

19 1.11 ORGANIZATION OF THE THESIS Chapter 1 presents the overview of MPLS and the basics of Traffic Engineering in MPLS. Chapter 2 highlights various existing methodologies proposed in the last few years. It discusses in detail about various congestion identification, avoidance and congestion control mechanisms. It also discusses the various Traffic Engineering methods. Chapter 3 explains the Balanced Traffic Distribution in MPLS using Gossip Based Aggregation Method. Chapter 4 illustrates the proposed Ant Colony Optimization method for congestion control in detail. Chapter 5 explains the traffic flow analysis model of MPLS networks. The flow analysis model categorizes the traffic depending on the flows and the path selection and, the implementation is discussed in detail. Chapter 6 demonstrates the proposed Hybrid Traffic Management Model, its design, algorithm and the performance analysis in detail. The conclusions and recommendations for the future enhancements are summed up in Chapter 7. 19