A high-performance OC-12/OC-48 queue design prototype for input-buffered ATM switches

Duan, Haoran; Lockwood, John W.; Kang, Sung-Mo; Will, J.D.

doi:10.1109/infcom.1997.635110

Cited by 40 publications

(22 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Switching matrix scheduling algorithms have been extensively studied for electronic switches [17]- [19]. However, they are similar to the burst-scheduling algorithms in the OBS routers only in a sense that the goals of the both types of scheduling algorithms are to obtain the switching matrix configurations such that input traffic can be efficiently sent to the desired outputs.…”

Section: Background On Scheduling Algorithmsmentioning

confidence: 99%

Optimal Burst Scheduling in Optical Burst Switched Networks

Chen

Turner

2007

J. Lightwave Technol.

View full text Add to dashboard Cite

Abstract-Optical burst switching (OBS) is an emerging technology that allows variable size data bursts to be transported directly over dense wavelength division multiplexing links. In order to make OBS a viable solution, the burst-scheduling algorithms need to be able to utilize the available wavelengths efficiently, while being able to operate fast enough to keep up with the burst incoming rate. For example, for a 16-port OBS router with 64 wavelengths per link, each operating at 10 Gb/s, we need to process one burst request every 78 ns in order to support an average burst length of 100 kB. When implemented in hardware, the well-known horizon scheduler has O(1) runtime for a practical number of wavelengths. Unfortunately, horizon scheduling cannot utilize the voids created by previously scheduled bursts, resulting in low bandwidth utilization. To date, minimum starting void is the fastest scheduling algorithm that can schedule wavelengths efficiently. However, while its complexity is O(log m), it requires 10 log m memory accesses to schedule a single burst. This means that it can take up to several microseconds for each burst request, which is still too slow to make it a practical solution for OBS deployment. In this paper, we propose an optimal burst scheduler using constant time burst resequencing (CTBR), which has O(1) runtime. The proposed CTBR scheduler is able to produce optimal burst schedules while having processing speed comparable to the horizon scheduler. The algorithm is well suited to highperformance hardware implementation.Index Terms-Algorithm, optical burst switching (OBS), optical packet switching, scheduling, wavelength division multiplexing (WDM), wavelength routing.

show abstract

Section: Background On Scheduling Algorithmsmentioning

confidence: 99%

Optimal Burst Scheduling in Optical Burst Switched Networks

Chen

Turner

2007

J. Lightwave Technol.

View full text Add to dashboard Cite

show abstract

“…A number of CM-SAs for IQ switch architectures have appeared in the technical literature; see, e.g., [2]- [4], [9] [11]- [16]. In this paper, we consider six proposals, namely, three optimal algorithms MWM-QL (MWM with queue lengths as weights), MWM-CA (MWM with cell ages as weights), MSM (maximum size matching), and three heuristics, iLQF [22], iSLIP [23], and iOCF [22].…”

Section: B Considered Cell-mode Scheduling Algorithms (Cm-sas)mentioning

confidence: 99%

“…Most of the implemented high-speed IQ switches internally operate on fixed-size data units (cells): the Lucent GRF [5], the Cisco GSR [6], the Tiny-Tera [7], the AN2/DEC [3], [8], the iPoint [9], and the MGR/BBN [10].…”

mentioning

confidence: 99%

See 1 more Smart Citation

Packet-mode scheduling in input-queued cell-based switches

Marsan

Bianco

Giaccone

et al. 2002

IEEE/ACM Trans. Networking

View full text Add to dashboard Cite

Abstract-We consider input-queued switch architectures dealing at their interfaces with variable-size packets, but internally operating on fixed-size cells. Packets are segmented into cells at input ports, transferred through the switching fabric, and reassembled at output ports. Cell transfers are controlled by a scheduling algorithm, which operates in packet-mode: all cells belonging to the same packet are transferred from inputs to outputs without interruption. We prove that input-queued switches using packet-mode scheduling can achieve 100% throughput, and we show by simulation that, depending on the packet size distribution, packet-mode scheduling may provide advantages over cell-mode scheduling.Index Terms-Input queued switched, packet switching, scheduling algorithms, variable size packets.

show abstract

“…Researchers have proposed several good switch scheduling algorithms, such as iterative SLIP (iSLIP), 7 iterative longest queue first (iLQF), 8 reservation with preemption and acknowledgment (RPA), 9 and matrix unit cell scheduler (MUCS) 10 With centralized implementations, the runtime of these algorithms is O(N 2 ) or more. But by adopting parallelism and pipelining (which means adding spatial complexity in hardware) these algorithms can considerably reduce their time complexity.…”

mentioning

confidence: 99%

An efficient randomized algorithm for input-queued switch scheduling

Shah

Giaccone²,

Prabhakar³

HOT 9 Interconnects. Symposium on High Performance Interconnects

View full text Add to dashboard Cite

Many networking problems suffer from the so-called curse of dimensionality: That is, although excellent (even optimal) solutions exist for these problems, they do not scale well to high speeds or large systems. In various other situations where deterministic algorithms' scalability is poor, randomized versions of the same algorithms are easier to implement and provide surprisingly good performance. For example, recent work in load balancing 1,2 and for documenting replacement in Web caches 3 provides compelling demonstrations of the effectiveness of these randomized algorithms. Motwani and Raghavan provide other examples and a good introduction to the theory of randomized algorithms. 4 Here, we focus on applying randomization to the design of input-queued (IQ) switch schedulers. We take for granted the effectiveness of the IQ architecture for very high-speed and for large-sized switches. Several references attribute this effectiveness to the IQ architecture's minimal memory bandwidth requirement compared with output-queued and shared-memory architectures. Figure 1 shows the logical structure of an N × N IQ packet switch. We assume the switch operates on fixed-size cells (or packets). Each input has N first-in first-out virtual output queues (VOQs), one for each output. This VOQ architecture avoids performance degradation from the head-of-the-line blocking phenomenon. 5In each time slot, at most one cell arrives at each input and at most one cell can transfer to an output. When a cell with destination output j arrives at input i, the switch stores it in the VOQ, denoted Q ij . Let the average cell arrival rate at input i for output j be λ ij . Incoming traffic is admissible if Σ i=1 N λ ij < 1, ∀ j; and Σ j=1 N λ ij < 1, ∀ i. In other words, these conditions ensure that no input or output is oversubscribed.We can model the scheduling problem as a matching problem in a bipartite graph with

show abstract

A high-performance OC-12/OC-48 queue design prototype for input-buffered ATM switches

Cited by 40 publications

References 27 publications

Optimal Burst Scheduling in Optical Burst Switched Networks

Optimal Burst Scheduling in Optical Burst Switched Networks

Packet-mode scheduling in input-queued cell-based switches

An efficient randomized algorithm for input-queued switch scheduling

Contact Info

Product

Resources

About