Distributed Scheduling in Input Queued Switches

Abstract

Dealing with RTTs (Round Trip Time) in IQ switches has been recently recognized as a challenging problem, especially if considering distributed (multi-chip) scheduler implementation which are suited to reduce the hardware complexity in very large, high-speed, switches. Traditional iterative three-or two-phase scheduling algorithms are based on a monolithic implementation, thus allowing instantaneous information exchange among input and output selectors to determine a matching. Multichip implementation imply that information exchange among inputs and outputs is delayed by an inter-chip latency. This delay requires non-trivial modifications to scheduling algorithms to allow a fully distributed implementation while keeping good performance. We propose a new scheduling algorithm, named SRR (Synchronous Round Robin), which is suited to a fully distributed implementation and provides good performance if compared with more complex, non fully distributed, previously proposed scheduling algorithms.

Feedback-Based Scheduling for Load-Balanced Two-Stage Switches

“…Therefore, SRR serves as a benchmark for single-stage input-queued switches. For more details of SRR, please refer to [27]. Note that we do not compare with LQF_Byte-Focal [10] and CR [16] because they cannot be used for multicabinet implementation.…”

“…• SRR (Synchronous Round-Robin) algorithm [27]. When the propagation delay is , we denote SRR as SRR/ .…”

An efficient single-iteration single-bit request scheduling algorithm for input-queued switches

“…Despite the fact that multi-iteration scheduling algorithm produces the maximal size match, it is not scalable (Mneimneh, 2008;Chen et al, 2009;Scicchitano et al, 2007).…”

“…SRR (Scicchitano et al, 2007) is a single-iteration single-bit request scheme. For any input port i, the preferred request to output port j is also calculated by a pre-determined and periodic sequence of N patterns.…”

“…The transmission delay and communication overhead also increase with the number of bits per request. For high speed routers/switches design, a single-iteration single-bit request approach (Scicchitano et al, 2007) is preferred. (For a detailed review on existing iterative scheduling algorithms, please refer to Section 3.…”