DonNTU   Masters' portal

Emjali saif

Department of Computer Science and Informatics

Department of Applied Mathematics and Computer Science

Specialty "Software Engineering"

Models for the analysis of flow control in computer networks CISCO

Scientific adviser: Associate Professor Dr. Ladyzhenskyy Y.V

Resume Abstract

Abstract

Summary on the theme of master's work

Contents

• Introduction
• 1. The organization and operation of computer networks
• 2. Computer network as a queuing system
• 3. Realny time scale
• 4. Quality functioning computer networks
• 5. Streams of service calls
• 5.1 Stationary and non-stationary flows of requests
• 5.2 simplest stream
• 6. Multiprotocol Label Switching
• Conclusion
• Source List

  Introduction

  Until now, the Internet service providers and large companies had to create and maintain separate networks for voice transmission, video, traffic needed for mission-critical, and the rest of the network traffic. However, it should be noted has developed in recent years a clear tendency to unite all of these networks into a single network packet data transmission protocol based on Internet Protocol (IP). The largest IP-sets is, of course, the global network Internet. Over the past few years, the growth of Internet, transferred over the network traffic and the number of existing applications Intene-close to exponential. While the Internet and corporate intranets continue to rise, many analysts are predicting the emergence of applications that focus on the transfer of non-traditional types of information, such as voice sets IP (Voice over IP - VoIP) and video conferencing traffic to pass. As the number of Internet users and a variety of network applications is increasing every day, the Network needs funds, which would provide support for both existing and emerging applications and services. However, to date, the Internet could provide only best-effort delivery of data (best effort service). Non-guaranteed delivery of data does not imply any kind of warranty relating to the time and the fact of the packet arrival at the destination. In this case, it should be noted that the packet drops can occur only at the time of network congestion. (A more detailed but non-guaranteed packet delivery is discussed in the section "Levels of quality of service," later in this chapter).[1]

  1. The organization and operation of computer networks

  Computer network (CN) is defined as the set of nodes (computers, terminals, peripherals) with the possibility of information exchange with each other using a special communication equipment and software. Dimensions networks vary widely - from a pair of interconnected computers, standing on a nearby table to millions of computers scattered around the world (some of which may be located in space objects). By the scale of the network can be divided into several categories. Local area network, LAN or LAN (Local-Area Network), allow you to combine computers located in a confined space. For local networks, as a rule, laid specialized cable system, and the position of the possible points of connection fees limited to this cable system. Sometimes used in local area networks and wireless communications (wireless), but also to the possibility of moving the subscribers is very limited.

  Ethernet can be combined into a large-scale education - CAN (Campus-Area Network) - a campus network of local networks of closely spaced buildings), MAN (Metropolitan-Area Network - a network of urban scale), WAN (Wide-Area Network - large-scale network), GAN (Global-Area Network - WAN). Network of networks nowadays called global network - Internet (Internet). For larger networks also established special wired or wireless connection, or use the existing infrastructure of public communication. In the latter case, the subscriber network can connect to the network in a relatively random locations covered by the network telephony, ISDN or cable TV.

  concept intranet (intranet) denotes an internal network of an organization where there are two important points:

  • isolation or protection of the internal network from the external (Internet);

  • Use IP network protocol, and Web-based technologies (Application protocol HTTP). In hardware aspect, the use of intranet technology means that all subscribers to basically communicate with one or more servers on which major information resources of the enterprise. In

  networks use different network technologies, of which the most common LAN Ethernet, Fast Ethernet, Gigabit Ethernet, Token Ring, l00VG-AnyLAN, ARCnet, FDDI. Network equipment is divided into active: PC interface cards, repeaters, hubs, etc. Passive cables, connectors, patch panels, etc. In addition, there is support equipment: uninterruptible power supply, air conditioning and accessories: mounting brackets , cabinets, cable trays of different types.

  computer networking equipment is divided into finite system (apparatus) is the source and / or users of information and intermediate systems providing information transport through the network. By the end systems (End Systems - ES), include computers, terminals, network printers, fax modems, POS, barcode readers, means of voice, video, and any other peripheral devices comprising one or another network interface. For intermediate systems, (Intermediate Systems - IS), are hubs (repeaters, bridges, switches), routers, modems and other telecommunications devices, as well as connecting them cabling and / or wireless infrastructure.

  main objective of the Network in terms of users, is the exchange of information between end devices. The data stream transmitted over the network is called a network traffic. Traffic except for useful information includes the service part of it - the inevitable overhead on the organization of interaction of network nodes. The capacity lines, also known as bandwidth (bandwidth), defined as the amount of data flowing through the line per unit time. It is measured in bits / sec (bps - bit per second), kbit / s (kbps), megabits / sec (Mbps), Gbit / s (Gbps), Tb / s (Tbps). Here, as a rule, prefixes kilo, mega, giga, tera, have a decimal value (103, 106, 109, 1012), rather than the binary (210, 220, 230, 240). For active communication equipment performance applicable concept, and in two different aspects. In total unstructured data skip equipment per unit time (bit / s), interest rate, and packet processing (pps - packets per second), frame (tps - frames per second) or cells (cps - cells per second). Naturally, at the same time saying and the size of the structures (blocks, frames, cells), which is measured by the speed of processing. Ideally, the performance of the communication equipment should be so high as to ensure the processing of data coming on all interfaces (ports) at their full speed (wire speed).

  Master's thesis is devoted to the actual scientific task of developing a unified approach to the synthesis of Moore automata directed to reduce hardware expenses in the resulting device, including algorithmic, combinatorial and schematics optimization techniques. The target of the basis used chip FPGA company Xilinx, combining functionality programmability, reconfigurability and wide availability of the consumer, and the tools of the study is CAD Xilinx ISE, Verilog HDL and Java SE.[2]

  2. Computer network as a queuing system

  In general it may be said that the computer network to solve a set of processing tasks streams service requests. Programs P1, ..., Pr solutions maintenance tasks are stored in permanent memory or other intellectual hub. They are initiated in the order determined by the processes taking place in the environment of the transmission network and the node itself. The reason for the initiation of treatment program is called. Claim. Requests are generated in the source object to the network and enter the node receiver periodically or at random, random times. In this short period of time can do several applications za, ..., zw, for which the service should be made relevant programs serving node Pa, ..., Pw. If only one processor in the service node, these programs can only be done consistently in this connection, there are queues for service requests. If you have multiple processors (like switches) queue for service (application for transmission of a packet to the destination port can be put in place) are formed in relation to employment destination port.

  Processing of applications to the Constitutional Court is organized by the circuit shown in Fig. Applications z1, ..., ZM enter the interrupt device that is part of a network node. When the application device zi interrupt operation initiates an interrupt to the host, as a result of which the node is switched to the program receiving and placing orders in a queue, called "Manager 1."

  

  Drawing - Processing of applications in the node Kc

  (animation: 4 frames, 4 cycles of repetition, 31 kilobytes)

  (Zi - input device, Oi - turn ,Пi - output device)

  «Manager 1" identifies the type of applications received and puts the request to the appropriate queue O. service. All in the physical aspect consists of a set of memory cells, which house the codes of applications received. In general, many of the queues contain requests awaiting service. Suppose that each time a node may perform only one service program. The process of selecting the application from a variety of applications waiting for service, called dispatching.

  scheduling procedure is implemented the program "Manager 2", which analyzes the state of the queues O1, ..., ON, select ZK application, the pre-emptive right to service and initiates the appropriate application PC. It is believed that the end of the program catered entity exits the system. At the end of the PC program control is transferred back "Dispatcher 2", which selects the next request and initiates the appropriate application. If the queues are empty, "Manager 2" switches the processor into sleep mode.

  Thus, the application process is carried out in two stages: the first stage of the application is accepted, "Manager 1" in the system and put in a queue to await service, in the second stage application selected "Manager 2", which organizes service requests, initiating appropriate application. Program "Manager 1" and "2 Manager" control the order of processing of applications and belong to the class of control programs. "Manager 1" is a short program, which is accomplished dozens of operations of the processor. It is initiated by a device interrupts at the moment of his entrance on a regular application. In this work the processor is busy executing a program is interrupted. "Manager 1", and the last job transfers control "Dispatcher 2", which continues the interrupted program or develop service applications received, if it has a preferential right to service compared to other applications.

  Rule scheduling, based on which of the queues are selected service requests, called service discipline. The specified service discipline implemented by the control program - "Manager 2".[3]

  3.Realny time scale

  A characteristic feature of these systems (computer network) - the presence of limiting restrictions on the stay application in the system. This time is equal to the sum of the waiting time when the application is waiting in queues, and service time spent on the execution of the corresponding node of the program application process. Mode, in which there are restrictions on the time limit of the programs in the node will be called work in real time (PMB). PMB provides the correct choice of the performance of devices that are part of the network planning and execution order of programs in which the most important tasks are implemented in the first place. Planning is reduced, in particular, to the initiation of programs in an order different from the order of receipt of applications to their execution. For example, if the system receives the application procedure za, zb, zg, zd, then the need to reduce the processing time of applications b and g of the program should be implemented in the following order: Pa, Pb, Pg, Pd.

  4.Quality functioning of computer networks

  The quality of the network is determined primarily by the quality of its components and the residence time in the application, characterized by the time interval from the moment of receipt of the application site to the end of its service, ie, until the time at which the application exits the system. Consequently, the definition of the residence time of applications in the queue - the primary task of analyzing the quality of the operation of computer networks. Functioning nodes CN can be seen as a combination of three processes:

  • receipt of applications to the input of the system;

  • scheduling applications;

  • Queuing. It is expected that the residence times of applications depend on the parameters of each of these processes.

  The nodes of the COP program processing requests for services performed in the order of time of receipt of applications, which are generally random. Also the timing of the same program are generally not constant and depends on the type of requests coming from outside the system and its condition. Thus, the operation of the COP is stochastic in nature, which allows us to consider such a system as a queuing system, and describe them in terms of queuing theory.[4]

  5. Flows of requests for service

  Set of events distributed in time, queuing theory called the flow of applications. Example data flow in Fig. 3.2. Here t1, t2, ..., - the moments of the event giving rise to the application. Distinguish between incoming and outgoing flows of applications which enter in and leave it, respectively. In general, the flow of applications is considered as a random process, defined by the distribution function between the time the receipt of two adjacent applications. A key characteristic of the flow - the intensity L, equal to the average number of requests received by the unit time. The value of 1 / L, the inverse of intensity determines the average time interval between two consecutive requests. Thus, if the intensity of L = 5 c1, the average interval between applications T = 1 / L = 0,2 s, if L = 0,2 then T c1 - 5 sec. [5][6]

  

  5.1 Stationary and non-stationary flows of requests

  The flow of applications may be stationary and non-stationary, if the flow characteristics do not change over time, it is called a stationary, non-stationary - if the characteristics change over time.

  Features node CN is most easily determined for steady-state operation of the system, assumed to be stationary flow applications. For this reason, time-dependent flows are approximated by certain time intervals stationary flows. For example, the flow of orders from customers arriving at the server, can not be fixed within a day, because the intensity of file transfers at the beginning and at the end of the day, usually much lower than during the day (see Fig.). From the graph shows that the highest intensity to the period from 9 to 15, within which a certain approximation the flow may be regarded as stationary.

  

  Drawing - The intensity of the flow of requests to a file server for the day

  5.2 simplest stream

  In queuing theory the largest number of analytical results obtained for the flow are called simple. The simplest flow has the following properties:

  • Attached (probabilistic characteristics of the flow does not depend on time);

  • lack of aftereffect (the requests come into the system independently of each other, in particular the length of the interval of time before the arrival of the next application does not depend on whether received at the initial application or not);

  • ordinariness (every time the system can do no more than one application). For the simplest flow intervals between successive applications - independent random variables with distribution function:

  

  distribution of this kind is called exponential (exponential) and has a density of

  

  
  

  Graphs densities corresponding to different values ??of parameter T, shown in Fig.

  

  Figure - Frequency distribution of time intervals between applications in the elementary stream

  expectation of the length of the time interval between successive moments of receipt of applications is defined as:

  

  dispersion time interval between successive moments of receipt of applications:

  

  calculate the probability of occurrence of short time intervals between two consecutive requests, the length of which is less than the expectation

  

  Thus, short intervals more frequent than long, ie, the simplest flow applications show a tendency to group together, which creates a more difficult conditions during operation node network compared to other distributions flow applications. For the simplest flow volume of requests to the system during the time interval t, the Poisson distribution:

  

  Fig. shows the shape of the Poisson distribution for some particular values. The expectation and variance of the Poisson distribution:

  

  

   

  6. Multiprotocol Label Switching

  Working Group on the creation and implementation of Multiprotocol Label Switching (Multiprotocol Label Switching (MPLS) Working Group) is standardizing core technology, involving the use of a forwarding mechanism based on the replacement of labels (label switching method) and is used in conjunction with the routing of the network layer. The purpose of the working group is the development of mechanisms for sharing technology multiprotocol label switching with various link layer technologies, including the Packet-over-Sonet, Frame Relay, ATM and Ethernet at data rates of 10, 100 and 1000 Mbit / s. The standard is based on the mechanism of MPLS switching tags (tag switching), developed by Cisco Systems.

  Among other advantages of MPLS technology allows greater flexibility in the provision of QoS and traffic management. To identify a traffic stream with certain QoS requirements using keywords which, inter alia, allow selection of optimized transmission path package. Technology MPLS, virtual private networks (VPNs) using MPLS technology and traffic management in MPLS networks have been applied mainly in service provider networks. More details multiprotocol label switching and quality of service in MPLS networks are discussed in Chapter 9, "The quality of service in networks MPLS", and manage traffic in the network MPLS - Chapter 10, "Managing traffic in networks MPLS".[8][9]

  Conclusions

  This project is designed to be a valuable source of information for network administrators, architects and engineers who want to learn and implement in their networks, services IP QoS. Quality of Service are urgently needed in today's scalable IP-based networks, providing secure and provide differentiated services Internet by transferring control of network resources and their use of the network operator.

  aim of this project is the architecture of the IP QoS mechanisms and their associated functions that implement quality of service through the corporate intranet, and service provider networks, in general, in the global network Internet. The main focus when considering the architecture of IP QoS mechanisms is on the architecture of differentiated services (diffserv). In addition, this book focuses on the mechanisms of QoS, provided technologies ATM, Frame Relay, IEEE 802.1p, IEEE 802.1Q, MPLS, and MPLS VPN, as well as the interaction of these mechanisms with IP QoS mechanisms to provide end to end service. Another no less important topic touched upon in this book is the theme of managing traffic in the networks of MPLS.

  The project is the most complete description of the IP QoS and all the associated quality of service technologies. Consideration of the ends of each technology a practical example, clearly illustrating its application. Readers of this book will receive detailed information on the following issues.

  • Fundamental principles of quality of service in IP networks and the need to introduce mechanisms for QoS.

  • Architecture differentiated services (diffserv) and permitting functions QoS.

  • Architecture of integrated services (Intserv) and permitting functions QoS.

  • Mechanisms for QoS, provided technologies ATM, Frame Relay, and IEEE 802.1p/802.1Q, - interaction with the mechanisms of IP QoS.

  • MPLS and MPLS VPN - interaction with the mechanisms of IP QoS.

  • Directing traffic in the networks of MPLS.

  • Policy routing, general IP QoS features and other information relating to the quality of service.

  QoS mechanisms may be implemented in any IP-based network. Therefore, the material presented in this book applies to all networks IP-corporate intranets, service provider networks, and, finally, Internet.[9][10]

  Source List

  1. Anthony Alles ATM Internetworking (born) / / Cisco Systems, Inc. - May 1995.

  2. Steve G. Steinberg Netheads vs Bellheads (born) / / Wired. - 1996. - № 4.10.

  3. A. Nazarov, IA Razzhivin, M. Simonov ATM: Technical solutions networking. - Reference book. - M.: Hot Line - Telecom, 2001. - S. 376. - ISBN 5-93517-040-X

  4. A. Nazarov, IA Razzhivin, M. Simonov ATM: Principles and technical networking solutions. - Textbook. aid for students studying in the field 200900 - "Communication networks and switching systems." - M.: Hot Line - Telecom, 2002. - S. 408. - ISBN 5-93517-079-5

  5. Galina Dicker-Pildush ATM Networks Corporation Cisco = Cisco ATM Solutions. - M.: "Williams", 2004. - S. 880. - ISBN 1-57870-213-5

  6. Technology Guide converged networks = Internetworking Technologies Handbook. - 4. - M.: "Williams", 2005. - S. 1040. - ISBN 5-58705-119-2

  7. lectures on the subject "Computer networks".

  8. electronic resource http://ru.wikipedia.org/wikiКомпьютерная_сеть.

  9. electronic resource http://www.quizful.net/post/ethernet-routing.

  10. Clark, Kevin Hamilton, "Principles of switching in LANs Cisco».

  Resume