Abstract

In this paper, we consider the performance of optical circuit switching (OCS) systems designed for data center networks by using network-level simulation. Recent proposals have used OCS in data center networks but the relatively slow switching times of OCS-MEMS switches (10–100 ms) and the latencies of control planes in these approaches have limited their use to the largest data center networks with workloads that last several seconds. Herein, we extend the applicability and generality of these studies by considering dynamically changing short-lived circuits in software-controlled OCS switches, using the faster switching technologies that are now available. The modelled switch architecture features fast optical switches in a single hop topology with a centralized, software-defined optical control plane. We model different workloads with various traffic aggregation parameters to investigate the performance of such designs across usage patterns. Our results show that, with suitable choices for the OCS system parameters, delay performance comparable to that of electrical data center networks can be obtained.

Highlights

  • In recent years we have witnesed an unprecedented growth in internet traffic due to the advent of new applications which are based on high performance computing (HPC) and cloud computing infrastructure in data centers

  • We evaluate the performance of an enhanced optical circuit switching (OCS), suitable for use in HPC and in data center networks, that makes use of the faster switching technologies that are available

  • Operational parameters of our system, the traffic aggregation time, can be tuned to better accommodate circuits of various durations and we investigate the effect of such tuning on OCS performance with various workloads

Read more

Summary

Introduction

In recent years we have witnesed an unprecedented growth in internet traffic due to the advent of new applications which are based on high performance computing (HPC) and cloud computing infrastructure in data centers. Multimedia gaming and streaming applications, website searching, scientific calculations/computations and distributed services are some examples of bandwidth-hungry applications, and their bandwidth demand is increasing exponentially. Today’s data centers comprise thousands of servers with high speed links and are interconnected via switches and routers. As the size and complexity of data center deployments increase, meeting their high bandwidth requirements is a challenge. Optical networks for data centers have gained significant attention over the last few years due to the potential and benefits of using optical components.

Objectives
Results
Conclusion
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call