Matching output queueing with a combined input output queued switch

Chuang, Shang-Tse; Goel, Ashish; McKeown, Nick; Prabhakar, Balaji

doi:10.1109/infcom.1999.751673

Cited by 126 publications

(16 citation statements)

References 24 publications

(19 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…There is a theoretical observation that it is possible to emulate purely Output-Queued (OQ) switches with Combined Input Output Queued (CIOQ) Switches running at approximately twice the line-rate, or namely with a "speedup" of two [7]. However, it is not easy to implement an electrical switch with speedup of two due to the limited IO port counts or insufficient memory bandwidth at high line rates (to speed up the line-rate by two, you need either double the clock frequency or double the number of IO wirings, neither of which is practical).…”

Section: Hybrid Optical Switching Architecturementioning

confidence: 99%

Experimental Demonstration of End-to-End Message Passing for HPC systems through a Hybrid Optical Switch

Yin

Proietti

et al. 2011

37th European Conference and Exposition on Optical Communications

View full text Add to dashboard Cite

show abstract

Section: Hybrid Optical Switching Architecturementioning

confidence: 99%

Experimental Demonstration of End-to-End Message Passing for HPC systems through a Hybrid Optical Switch

Yin

Proietti

et al. 2011

37th European Conference and Exposition on Optical Communications

View full text Add to dashboard Cite

show abstract

“…We also need to define the Deadline To Leave(DTL) of the kth packet: DTLj (k) = TTLj (k) + ODB (2) Moreover, the memory bank of BC-DSM is organized similar to Virtual-Ouput-Queueing (VOQ) architecture in IQ [3]. The memory is virtually divided into N FIFO queues according to the output port number.…”

Section: Scheduling Algorithms Designmentioning

confidence: 99%

“…The first is CIOQ approach and the scheduling is based on GBVOQ ( [3]etc). This classical scheme runs the switch fabric at the speedup* of two and access memory at the speedup of three.…”

Section: Introductionmentioning

confidence: 99%

Emulating FCFS OQ with Buffered Crossbar Based Distributed Shared Memory Switch

Hua

Liu

et al. 2007

2007 Workshop on High Performance Switching and Routing

View full text Add to dashboard Cite

Emulating Output Queued (OQ) Switch with sustainable implementation cost and low fixed delay is always preferable in high performance router design. Distributed Shared Memory (DSM) Switch, also called Switch-MemorySwitch (SMS) router, provides a promising way towards emulating OQ in backbone switches. However, the previous architectures and algorithms for DSM/SMS switches require either high implementation cost or iterations of scheduling. This paper proposes a Buffered Crossbar based DSM switch (BC-DSM), which can emulate the FCFS OQ based on a speedup-two buffered crossbar. Each crosspoint has only two packet buffers, one for arrival packet and one for departure packet. Different with previous works of DSM, we do not aim to resolve memory access conflicts when packets arrive, but to randomly dispatch packets into memories through buffered crossbar and resolve departure conflicts by bounded small queueing and delay. For scheduling, the departure packets are prioritized in the transferring between the fabric and the memory.The advantages of our scheme include: 1) Only N two-speedup memories and one two-speedup buffered N x N crossbar are needed. 2) Scheduling process is simple and distributive, thanks to buffered crossbar. 3) No need for additional crosspoint buffer for overcoming long Round-Trip Time (RTT), thanks to the probabilistic load-balancing scheme. 4) Simulation shows that only a few fixed delay is needed to provide extremely low packet loss rate such as 10-8, under different types of traffic patterns, including bursty model and unbalanced model.

show abstract

“…In comparison Input Queued (IQ) switching fabric only need to work in line speed, but the integrated scheduling algorithms make it complex to realize QoS ( [4] [8] [9]). In this situation some researchers proposed a combined input and output queued (CIOQ) switching fabric and proved that CIOQ fabric can simulate OQ fabric at a speed up of 2 ([10] [11] [12]). However this fabric can not reach 100% throughput in multicast scenarios even when fanout splitting is allowed.…”

Section: Introductionmentioning

confidence: 99%

Practical network coding approach for multicast packet switching

Cheng

Wang

et al. 2009

2009 Global Information Infrastructure Symposium

View full text Add to dashboard Cite

This paper considers the problem of applying network coding in Input and Output Queued (IOQ) packet switch loaded with both unicast and multicast flows. With network coding the switch can transmit the linear combination of multicast packets which prominently improves the switching efficiency in terms of throughput. The key idea is to use Vandermonde Matrix as the coding coefficient matrix while applying linear network coding over multicast packets. The paper also proposes a practical NC-based switch model to implement the coding approach.

show abstract

Matching output queueing with a combined input output queued switch

Abstract: -The Internet is facing two problems simultaneously: there is a need for a faster switching/routing infrastructure, and a need to introduce guaranteed qualities of service (QoS

Cited by 126 publications

References 24 publications

Experimental Demonstration of End-to-End Message Passing for HPC systems through a Hybrid Optical Switch

Experimental Demonstration of End-to-End Message Passing for HPC systems through a Hybrid Optical Switch

Emulating FCFS OQ with Buffered Crossbar Based Distributed Shared Memory Switch

Practical network coding approach for multicast packet switching

Contact Info

Product

Resources

About