Survey of Performance Analysis on Banyan Networks Written By Nathan D. Truhan Kent State University.

Slides:



Advertisements
Similar presentations
IP Router Architectures. Outline Basic IP Router Functionalities IP Router Architectures.
Advertisements

Interconnection Networks: Flow Control and Microarchitecture.
CCNA3: Switching Basics and Intermediate Routing v3.0 CISCO NETWORKING ACADEMY PROGRAM Switching Concepts Introduction to Ethernet/802.3 LANs Introduction.
ATM Switch Architectures
Lecture 4. Topics covered in last lecture Multistage Switching (Clos Network) Architecture of Clos Network Routing in Clos Network Blocking Rearranging.
Delivery and Forwarding of
1 CNPA B Nasser S. Abouzakhar Queuing Disciplines Week 8 – Lecture 2 16 th November, 2009.
1 ELEN 602 Lecture 18 Packet switches Traffic Management.
Priority Scheduling and Buffer Management for ATM Traffic Shaping Authors: Todd Lizambri, Fernando Duran and Shukri Wakid Present: Hongming Wu.
1 Omega Network The omega network is another example of a banyan multistage interconnection network that can be used as a switch fabric The omega differs.
Router Architecture : Building high-performance routers Ian Pratt
1 Delta Network The delta network is one example of a multistage interconnection network that can be used as a switch fabric The delta network is an example.
What's inside a router? We have yet to consider the switching function of a router - the actual transfer of datagrams from a router's incoming links to.
Chapter 8 Switching Switching at the physical layer in the traditional telephone network uses the circuit-switching approach.
1 Performance Results The following are some graphical performance results out of the literature for different ATM switch designs and configurations For.
10 - Network Layer. Network layer r transport segment from sending to receiving host r on sending side encapsulates segments into datagrams r on rcving.
1 Advances in Optical Switching Mohammad Ilyas, PhD College of Engineering & Computer Science Florida Atlantic University Boca Raton, Florida USA
Chapter 10 Switching Fabrics. Outline Physical Interconnection Physical box with backplane Individual blades plug into backplane slots Each blade contains.
048866: Packet Switch Architectures Dr. Isaac Keslassy Electrical Engineering, Technion The.
048866: Packet Switch Architectures Dr. Isaac Keslassy Electrical Engineering, Technion Scaling.
ATM COMPONENTS Presented by: ANG BEE KEEWET CHONG SIT MEIWET LAI YIN LENGWET LEE SEANG LEIWET
Quickest path and Quickest routing: A dynamic routing method Research Topic: Jiang, XidongMS candidate in computer science at California State University,
8.1 Chapter 8 Switching Copyright © The McGraw-Hill Companies, Inc. Permission required for reproduction or display.
Dragonfly Topology and Routing
Gursharan Singh Tatla Transport Layer 16-May
(part 3).  Switches, also known as switching hubs, have become an increasingly important part of our networking today, because when working with hubs,
Pipelined Two Step Iterative Matching Algorithms for CIOQ Crossbar Switches Deng Pan and Yuanyuan Yang State University of New York, Stony Brook.
Buffer Management for Shared- Memory ATM Switches Written By: Mutlu Apraci John A.Copelan Georgia Institute of Technology Presented By: Yan Huang.
Switching Techniques Student: Blidaru Catalina Elena.
SIGNALING. To establish a telephone call, a series of signaling messages must be exchanged. There are two basic types of signal exchanges: (1) between.
Dynamic Networks CS 213, LECTURE 15 L.N. Bhuyan CS258 S99.
ATM SWITCHING. SWITCHING A Switch is a network element that transfer packet from Input port to output port. A Switch is a network element that transfer.
TO p. 1 Spring 2006 EE 5304/EETS 7304 Internet Protocols Tom Oh Dept of Electrical Engineering Lecture 9 Routers, switches.
1 Copyright © Monash University ATM Switch Design Philip Branch Centre for Telecommunications and Information Engineering (CTIE) Monash University
CHAPTER 12 INTRODUCTION TO PARALLEL PROCESSING CS 147 Guy Wong page
Multiplexing FDM & TDM. Multiplexing When two communicating nodes are connected through a media, it generally happens that bandwidth of media is several.
Computer Networks: Switching and Queuing Ivan Marsic Rutgers University Chapter 4 – Switching and Queuing Delay Models.
ISLIP Switch Scheduler Ali Mohammad Zareh Bidoki April 2002.
Case Study: The Abacus Switch CS Goals and Considerations Handles cell relay (fixed-size packets) Can be modified to handle variable-sized packets.
Shared Memory switches Masoud Sabaei Assistant professor Computer Engineering and Information Technology Department, Amirkabir University of Technology.
1 Data Link Layer Lecture 23 Imran Ahmed University of Management & Technology.
Jon Turner Resilient Cell Resequencing in Terabit Routers.
Forwarding.
Ch 8. Switching. Switch  Devices that interconnected with each other  Connecting all nodes (like mesh network) is not cost-effective  Some topology.
Buffered Crossbars With Performance Guarantees Shang-Tse (Da) Chuang Cisco Systems EE384Y Thursday, April 27, 2006.
1 IEX8175 RF Electronics Avo Ots telekommunikatsiooni õppetool, TTÜ raadio- ja sidetehnika inst.
110 February 2006 RapidIO FT Research Update: Dynamic Routing David Bueno February 10, 2006 HCS Research Laboratory Dept. of Electrical and Computer Engineering.
1 Buffering Strategies in ATM Switches Carey Williamson Department of Computer Science University of Calgary.
CCNA3 Module 4 Brierley Module 4. CCNA3 Module 4 Brierley Topics LAN congestion and its effect on network performance Advantages of LAN segmentation in.
1 Transport Layer: Basics Outline Intro to transport UDP Congestion control basics.
Spring 2000CS 4611 Router Construction Outline Switched Fabrics IP Routers Extensible (Active) Routers.
CS 4594 Broadband Switching Elements and Fabrics.
Input buffered switches (1)
Univ. of TehranIntroduction to Computer Network1 An Introduction to Computer Networks University of Tehran Dept. of EE and Computer Engineering By: Dr.
Network layer (addendum) Slides adapted from material by Nick McKeown and Kevin Lai.
McGraw-Hill©The McGraw-Hill Companies, Inc., 2000 Muhammad Waseem Iqbal Lecture # 20 Data Communication.
Chapter 3 Part 3 Switching and Bridging
Switching Techniques In large networks there might be multiple paths linking sender and receiver. Information may be switched as it travels through various.
Chapter 3 Part 3 Switching and Bridging
Packet Switching (basics)
Multiprocessors Interconnection Networks
Router Construction Outline Switched Fabrics IP Routers
Delta Network The delta network is one example of a multistage interconnection network that can be used as a switch fabric The delta network is an example.
EE 122: Lecture 7 Ion Stoica September 18, 2001.
Lecture: Interconnection Networks
Chapter 3 Part 3 Switching and Bridging
Introduction to Packet Scheduling
Switching Chapter 2 Slides Prepared By: -
Switch Performance Analysis and Design Improvements
Introduction to Packet Scheduling
Presentation transcript:

Survey of Performance Analysis on Banyan Networks Written By Nathan D. Truhan Kent State University

Outline Introduction & Banyan Networks Replicated & Dilated Banyan Networks Plane Interconnected Parallel Network Buffered Banyan Network Output-buffered Banyan Network Input-buffered Banyan Network Conclusions References

Introduction & Banyan Networks Banyan networks have been a topic of discussion in high-speed networks due to its low cost and simple implementation Belongs to a family of multistage interconnection networks which also include Omega, Shuffle-exchange, and others

Introduction & Banyan Networks Has the properties of having a unique path from source to destination, constructed out of m-stages of uniform switches, self- routing, modular, having a constant delay for all input-output port pairs, in-order delivery of packets

Introduction & Banyan Networks Consists of O(N log N) switching elements, split into O(log N) stages Standard Banyan has no buffers Will concentrate on 2 x 2 switches [1]

Replicated & Dilated Banyan Networks Expands a Banyan network by adding switching elements or links A Replicated Banyan network adds d copies of the Banyan network connected by a multiplexer and de-multiplexer

Replicated & Dilated Banyan Networks When a packet enters the network at the input port I, a layer in d is chosen and the packet is routed to this layer Layers in a replicated Banyan network are selected in one of three methods - Random, round robin and network load

Replicated & Dilated Banyan Networks [2]

Replicated & Dilated Banyan Networks Once a packet has traversed its assigned layer it reaches the output de-multiplexer The de-multiplexers can take packets in two ways - Multiple and single acceptance One adverse side-effect is that the Banyan network could lose its in-order delivery of packets

Replicated & Dilated Banyan Networks Can force all packets in same connection to same layer Can be more expensive to deploy, having an O(d x (N log N)), where d is number of layers in the network

Replicated & Dilated Banyan Networks Can compromise between standard Banyan and replicated Banyan by using partially replicated networks Replication occurs only in layers that see congestion

Replicated & Dilated Banyan Networks [2]

Replicated & Dilated Banyan Networks Throughput of replicated Banyan network [2]

Replicated & Dilated Banyan Networks Delay of replicated Banyan network [2]

Replicated & Dilated Banyan Networks A Dilated Banyan network adds d lines to each port to allow multiple packets to flow between switching elements at the same time which results in 2d x 2d switching elements It has also been shown in Architecture and Performance Analysis of DIRSMIN [8] that to meet the low-latency and high performance requirement, d = log log N dilation is optimal.

Replicated & Dilated Banyan Networks [3]

Plane Interconnected Parallel Network Introduced by Moustafa Youssef, Mohamed El- Derini and Hussien Aly from Alexandria University [4] Standard PIPN network created as split Banyan network with two sub-networks of size N/2 x N/2 and n – 1 stages Non uniform traffic throughput is measured with [4]:

Plane Interconnected Parallel Network [4]

Plane Interconnected Parallel Network Can use features of replication and dilation of Banyan networks [4]

Plane Interconnected Parallel Network [4]

Plane Interconnected Parallel Network Performance in a dilated PIPN network can be shown with dilation 2 and 4 [4]

Outline Introduction & Banyan Networks Replicated & Dilated Banyan Networks Plane Interconnected Parallel Network Buffered Banyan Network Output-buffered Banyan Network Input-buffered Banyan Network Conclusions References

Buffered Banyan Network Most common method to deal with contention is add a queue to some location in the switching element [5]

Buffered Banyan Network Queues can be placed at the input or output buffers and will have an impact on the performance on the network [5]

Buffered Banyan Network Cannot completely remove contention, since if packets are headed to the same queue we still have contention Contended packets are not dropped, but held in the buffer until it can be transmitted, unless the buffer is full

Buffered Banyan Network [7]

Buffered Banyan Network Processing a network buffer occurs in a two stage clock cycle At time 0, the network is empty In the first cycle the network checks for availability of free queue slot to transfer the head-of-line slot packet In the second cycle the routing of the HOL packet occurs

Buffered Banyan Network Buffers based off of a three-stage Markovian model [6] to show state of head-of-line packet –State “e” is empty, no packets in buffer –State “n” denotes new packet which is different from previous cycle –State “b” denotes packet in buffer is same as last cycle and was blocked

Buffered Banyan Network At the beginning of an operation cycle a buffer in stage k, may be in (2B + 1) possible combinations of states, where, and B is the number of buffers in the queue: [5] The last stage, m, is special as it cannot be in state b, therefore it only has (B + 1) states, giving us the following: [5]

Buffered Banyan Network if two packets are contending for the same input of the switching element in the next stage, but are destined for two separate outputs, with input buffering they would contend and cause congestion however with output queuing they would be permitted to enter the next stage and be placed in their separate output queues

Output-buffered Banyan Network Output-buffered Banyan has extra buffer stage at the inputs, while the rest reside at the outputs of the switching elements This leads to having m + 1 stages considering queuing

Output-buffered Banyan Network Performance is modeled similar to dilated PIPN network [5]

Input-buffered Banyan Network Unlike output-buffered network, all queues are at the inputs This leads to having exactly m stages

Input-buffered Banyan Network [6]

Outline Introduction & Banyan Networks Replicated & Dilated Banyan Networks Plane Interconnected Parallel Network Buffered Banyan Network Output-buffered Banyan Network Input-buffered Banyan Network Conclusions Questions References

Conclusions In this survey paper we have looked at several methods to improve a Banyan network Dilating and replication provide marginal increases as well as the PIPN network which combines these methods. The main problem with these designs is the lack of buffer, which still causes contention to occur and packets to be lost This contention problem is greatly reduced by adding buffers to the network, be it one buffer slot or multiple. There is not a foolproof way to eliminate contention because adding buffers may also add congestion All methods shown provide a marginal difference in performance, and even through they take different approaches to enhancing performance they are have all similar results, as demonstrated in the figures above

Questions?

References [1]C. Bouras, C. Gkantsidis. Cost of implementing Banyan networks for use in ATM switching fabrics. tingBanyanNetworksForUseInATMSwitchingFabrics.pdf [2]D. Tutsch, G. Hommel. Multilayer Multistage Interconnection Networks [3]M. Youssef, M. El-Derini, H. Aly. Performance Enhancement Techniques of a Banyan Network Based Interconnection Structure [4]A. Somani, T. Zhang. Architecture and Performance Analysis of DIRSMIN: A Fault-Tolerant Switch using Dilated Reduced-Stage MIN

References [5]K. Chan, et. al. A Refined Model for Performance Analysis of Output- buffered Banyan Networks.. [6]K. Chan, K. Yeung, S. Chan. A Refined Model for Performance Analysis of Buffered Banyan Networks with and without Priority Control. In IEICE Transactions, January 1999, Vol.E82-B.. [7]D. Koppelman. Congested Banyan Network Analysis Using Congested- Queue States amd Neighboring-Queue Effects