Metropolitan Area Networks

With the continuing success of Local Area Networks (IANs), there is an increasing demand to extend their capabilities towards higher data rates and wider areas. This, together with the progress in fiber-optic technology, has given rise to the so-called Metropolitan Area Networks (MANs). MANs can span much greater distances than current LAN s, and offer data rates on the order of hundreds of Megabits/sec (Mbps). The success of MANs is mainly due to the opportunity they provide to develop new networking products capable of providing high-speed commu­ nications between applications at competitive prices, which nonetheless give an adequate return on the manufacturers' investments. A major factor in of appropriate networking standards. achieving this goal is the availability Fiber Distributed Data Interface (FDDl) and Distributed Queue Dual Bus (DQDB) are the two standard technologies for MANs for which industrial products are already available. For this reason, this book focuses mainly on these two standards. Nowadays there are several books dealing with MANs, and these look mainly at FDDI (e.g., [2], [92], [118], [141]). These books focus primarily on the architectures and protocols, whereas they pay little attention to per­ formance analysis. Due to the capability of MANs to integrate services, a quantitative analysis of the Quality of Service (QoS) provided by these tech­ nologies is a relevant issue, and is thus covered in depth in this book.

1 Introduction.- 1.1 Capacity.- 1.2 From LANs to MANs.- 1.3 Fairness.- 1.4 User-oriented Performance Figures.- 1.5 Modeling of a MAN Environment.- 1.6 Outline of Contents.- 2 Stochastic Processes for Modeling Metropolitan Area Networks: Basic Results.- 2.1 Queueing Models for MAN Modeling.- 2.1.1 Single-queue Models.- 2.1.2 Single-queue Models with Server Vacations.- 2.1.3 Polling Models.- 2.2 System Properties.- 2.2.1 Little Theorem.- 2.2.2 Relationships between Arrival and Departure Distributions.- 2.2.3 Relationships between Arrival and Steady-state Distributions.- 2.3 Some Results on Renewal and Regenerative Processes.- 2.4 Work Conservation Laws.- 2.5 Stochastic Decomposition Laws.- 2.5.1 Stochastic Decomposition Laws for Single-queue Models.- 2.5.2 From the Decomposition Law to the Average Response Time.- 2.5.3 Stochastic Decomposition Laws for Polling Models.- 2.6 Pseudo-conservation Laws.- 2.6.1 Exhaustive Service Discipline.- 2.6.2 Gated Service Discipline.- 2.6.3 1-limited Service Discipline.- 2.6.4 l-limited Service Disciplines.- 3 Methods for the Analysis of Node-In-Isolation Models.- 3.1 M/G/l Systems.- 3.1.1 Infinite Buffer Systems.- 3.1.2 Finite Buffer Systems.- 3.2 M/G/l Systems with Vacation.- 3.2.1 Queue with Vacation and E-limited Service Discipline.- 3.2.2 Queue with Vacation and E-limited Service Disciplines.- 3.3 M/G/l-type Models.- 3.3.1 Solution Method.- 3.3.2 Implementing the Solution Method.- 3.3.3 Special Cases.- 3.3.4 Case Study: Analysis of a Packet-switching Node.- 3.4 Application of Node-in-isolation Models: Worst-case Analysis.- 3.4.1 Accuracy of the Worst-case Model Approach.- 3.4.2 E-limited Service Discipline.- 3.4.3 G-limited Service Discipline.- 4 Methods for the Analysis of Network-Wide Models: Polling Models.- 4.1 From Pseudo-conservation Laws to Waiting Time Analysis.- 4.1.1 Cycle Length Analysis: 1-limited Polling System.- 4.1.2 Cycle Length Analysis: l-limited Polling System.- 4.1.3 Waiting Time Analysis for 1-limited Polling Systems.- 4.1.4 Waiting Time Analysis for Polling Systems with Mixed Polling Strategies.- 4.1.5 Waiting Time Analysis for l-limited Polling Systems.- 4.2 Numerical Methods.- 4.2.1 Analysis of Finite Capacity Systems.- 4.2.2 Power Series Algorithm.- 4.3 For Further Study.- 4.3.1 Cycle Length Analysis: Convergence of the Iterative Procedure.- 4.3.2 1-limited Polling System: Details of the Computation.- 4.3.3 l-limited Polling System: Details of the Computation.- 4.3.4 Two-Moment Approximation of a Distribution.- 5 Fiber-Distributed Data Interface (FDDI).- 5.1 Introduction.- 5.2 FDDI Layers and Services.- 5.3 MAC Protocol.- 5.3.1 Token Structure.- 5.3.2 Frame Structure.- 5.3.3 FDDI Token Passing Mechanism.- 5.3.4 FDDI Timed Token Protocol.- 5.3.5 Additional FDDI Features.- 5.3.6 Claim Token Process.- 5.3.7 Beacon Process.- 5.3.8 Examples of Parameter Calculations.- 5.4 FDDI MAC Protocol Capacity.- 5.5 FDDI Cycle Properties.- 5.5.1 Maximum Cycle Length.- 5.5.2 Average Cycle Length.- 5.6 Remarks on the IEEE 802.4 Token Bus Protocol.- 5.7 Current Use of FDDI.- 6 FDDI Models.- 6.1 Introduction.- 6.2 Network-wide Models.- 6.2.1 Single-buffer Model with Synchronous and Asynchronous Traffic (Model 1).- 6.2.2 An FDDI Model with Zero Switchover Time (Model 2).- 6.2.3 An FDDI Network with Synchronous Traffic (Model 3).- 6.2.4 An FDDI Network with Asynchronous Traffic (Model 4).- 6.3 Station-in-isolation Models.- 6.3.1 M/G/l with Vacation and Exhaustive Limited with Limit Variation Service Discipline (Model 5).- 6.3.2 Worst Case Model for Synchronous Traffic (Model 6).- 6.3.3 M/G/l with Vacation and Vacation-dependent Time-limited Service Discipline (Model 7).- 6.3.4 M/G/l with Vacations and Time-controlled Service Discipline with and without Accumulated Delay (Model 8).- 6.4 Model 4: Details of the Computation.- 7 Distributed Queue Dual Bus (DQDB).- 7.1 Functional Architecture of a Node.- 7.2 Connectionless Data Service.- 7.2.1 MAC Convergence Function.- 7.2.2 DQDB MAC Protocol.- 7.2.3 DQDB MAC Protocol with Priorities.- 7.3 DQDB Performance and Fairness.- 7.3.1 Underload Analysis.- 7.3.2 Asymptotic Analysis.- 7.4 The Bandwidth Balancing Mechanism.- 7.4.1 Performance of DQDB with one Level of Priority and the BWB Mechanism.- 7.4.2 Performance of DQDB with Several Levels of Priority and the BWB Mechanism.- 7.5 DQDB MAC Protocol Capacity.- 7.6 Current Use of DQDB.- 7.6.1 The Tuscany MAN Testbed.- 8 DQDB Models.- 8.1 Introduction.- 8.2 Network-wide Models.- 8.2.1 Node-spaced Models.- 8.2.2 Node-concentrated Models.- 8.3 Node-in-isolation Models.- 8.3.1 L_NET Modeling.- 8.3.2 Tagged Node Models.- 9 Evolution Towards Gigabit Rates.- 9.1 Shared Medium Gigabit Networks.- 9.1.1 Cyclic Reservation Multiple Access (CRMA).- 9.1.2 MetaRing MAC Protocol.- 9.2 ATM-based Gigabit Networks.- 9.2.1 ATM LAN.- Acronyms.- Glossary of Notation.