Dynamic photonic lightpaths in the StarPlane network

https://doi.org/10.1016/j.future.2008.07.008Get rights and content

Abstract

The StarPlane project enables users to dynamically control network photonic paths. Applications running on the Distributed ASCI Supercomputer (DAS-3) can manipulate wavelengths in the Dutch research and education network SURFnet6. The goal is to achieve fast switching times so that when the computational pattern in the computing clusters changes, the underlying network topology adapts rapidly to the new traffic flow. StarPlane develops: the software to perform optimal traffic engineering; the management plane to map the users’ request into directives for the network control plane, as well as the integration to the DAS-3 Grid middleware. We present here some preliminary results obtained with few selected Grid applications that make use of StarPlane.

Introduction

StarPlane is a research project funded by the Netherlands Organization for Scientific Research (NWO) and carried out at the University of Amsterdam (UvA) and the Vrije Universiteit Amsterdam (VU). The goal of the project is to build an application-controlled photonic network. Nowadays the e-Science community relies on distributed collaborations with researchers interacting remotely with each other and exchanging large dataset [1], [2]. To meet these requirements, the use of dedicated and controllable network connections, a.k.a. lightpaths has increased [3].

To date, there are many projects working at making lightpaths a reality. The OptIPuter project [4] looks at creating the global tools and infrastructure that allow scientists to cooperate efficiently over optical networks. The International Center for Advanced Internet Research (iCAIR) [5] proposes ODIN, the Optical Dynamic Intelligent Network [6]; this is a control plane architecture that provides deterministic lightpath service as a capability complementary to traditional packet routing. The OSCARS [7] project investigates network resource reservation at the intra- and inter-domain level and provisions MPLS (Multi-Protocol Label Switching) end-to-end virtual circuits over a Layer3 IP network; whereas the DRAGON [8] project uses Generalized MPLS (GMPLS) to set up lightpaths over a Layer1 lambda switched network. The EU-funded project Phosphorus [9] mainly focuses on the key technical challenges to enable on-demand e2e network services across multiple domains. Phosphorus will make applications aware of their complete Grid resources (computational and networking) environment and capabilities, and able to make dynamic, adaptive and optimized use of heterogeneous network infrastructures connecting various high-end resources.

The distinctive features of StarPlane in comparison with these other initiatives are the accent on fast reconfiguration times, ideally in the orders of seconds or sub-seconds, the use of photonic equipment in the core, and the tight coupling between the user application and the networking services offered.

The rest of the article is organized as follows. Section 2 outlines the network architecture of StarPlane and describes the technical challenges related to the setup of the full end-to-end paths; Section 3 introduces the middlewares aspect and most specifically the management plane of StarPlane; Section 4 gives details on some of the most promising applications currently using StarPlane.

Section snippets

The StarPlane network

As the StarPlane network architecture has been described in detail in previous publications [10], in this section we cover only the photonic core of the network and we explain what are the engineering challenges of the project.

CPL (Common Photonic Layer) is the Nortel WDM (wavelength-division multiplexing) equipment used in SURFnet6 to carry DWDM (Dense WDM) signals on optical fibers; SURFnet deploys optical multiplexers, optical amplifiers and the eROADM (Enhanced Reconfigurable Optical

StarPlane middlewares

The StarPlane middleware has three main components:–the Dynamic Resource Allocation Controller (DRAC), developed by Nortel Networks, that provides the photonic switching service;–the StarPlane Management Plane (SPMP) that builds new services used by the higher level middlewares and user-applications; - and the Grid execution service, KOALA,that is in charge of the overall orchestration.

Applications on StarPlane

We will now take a closer look at two applications, Awari and SCARIe, currently using StarPlane. In particular SCARIe, with its large amount of data to be retrieved from remote locations via optical networks and to be correlated centrally, is an example of the new challenges projects like OptIPuter and StarPlane try to solve.

Awari is an application that solves an ancient two-player board game using retrograde analysis. This board game is played using 48 stones divided over 12 pits

Future work and conclusions

The preliminary results we obtained from monitoring tools and user tests using the Enhanced Socket API (see Section 3.4) show unexplained behaviors of the throughput in the StarPlane network. Our first goal is to investigate the origin of the oscillations visible in Fig. 2. We suspect this is due to unexpected interactions between cluster nodes, the Myrinet switches and the OMEs.

In Section 3.2 we described our current work on the use of logical programming (queries) for the interaction with the

Acknowledgments

We wish to thank SURFnet for their continuous support, in particular Roeland Nuijts, Bram Peeters and Antoon Prins. We also thank Nico Kruithof and the rest of the SCARIe project. The StarPlane research is supported by SURFnet, the NWO grant 643.000.504 and the BSIK program GigaPort.

Paola Grosso received her Ph.D. in Physics from the University of Turin - Italy. She worked at the Stanford Linear Accelerator Center and since 2004 she is a member of the SNE (System and Network Engineering) group of the University of Amsterdam. She is leading the group effort in the fields of optical networking and her current research interests are lightpath provisioning and resource scheduling, and semantic network models. Dr. Grosso is chair of the OGF NML-WG, Network Markup Language

References (22)

  • Robert L. Grossman et al.

    Experimental studies using photonic data services at IGrid 2002

    Future Generation Computer Systems

    (2003)
  • Larry Smarr et al.

    Special section: iGrid2005: The Global Lambda Integrated Facility

    Future Generation Computer Systems

    (2006)
  • Cees de Laat et al.

    The rationale of the current optical networking initiatives

    Future Generation Computer Systems

    (2003)
  • The OptIPuter project,...
  • The iCAIR website,...
  • J. Mambretti et al.

    Optical dynamic intelligent network services (ODIN): An experimental control-plane architecture for high-performance distributed environments based on dynamic lightpath provisioning

    IEEE Communications Magazine

    (2006)
  • Chin P. Guok, David Robertson, Mary Thompson, Jason Lee, Brian Tierney, William Johnston, Intra and interdomain circuit...
  • Tom Lehman, Xi Yang, Chin P. Guok, Nageswara S.V. Rao, Andy Lake, John Vollbrecht, Nasir Ghani, Control plane...
  • S. Figuerola, N. Ciulli, M.De Leenheer, Y. Demchenko, W. Ziegler, A. Binczewski, PHOSPHORUS: Single-step on-demand...
  • Paola Grosso et al.

    StarPlane - a national dynamic photonic network controlled by grid applications

    Emerald Journal of Internet Research

    (2007)
  • Patrick Blackburn et al.
  • Cited by (0)

    Paola Grosso received her Ph.D. in Physics from the University of Turin - Italy. She worked at the Stanford Linear Accelerator Center and since 2004 she is a member of the SNE (System and Network Engineering) group of the University of Amsterdam. She is leading the group effort in the fields of optical networking and her current research interests are lightpath provisioning and resource scheduling, and semantic network models. Dr. Grosso is chair of the OGF NML-WG, Network Markup Language Working Group.

    Damien Marchal, University of Amsterdam Damien Marchal received his Ph.D. from the University of Lille (France) in the field of computer graphics and computer animation. He is now working as Post-doc on the SCARIe project focusing on interaction between SCARIe and StarPlane. His research interests include field interpolation techniques and discreet differential geometry as well as logic-based semantic reasoning and high performance distributed computing.

    Jason Maassen obtained his Ph.D. at the VU University Amsterdam in 2003 on “Method Invocation Based Programming Models for Parallel Programming in Java”. He is now working as a postdoc at the same University. In the past, his research has focused on programming models, communication libraries and middleware for Grid computing. His current research is part of the StarPlane Project, and examines how grid applications can benefit from using a reconfigurable photonic network.

    Eric Bernier is an engineer in the CTO Office at Nortel where he works on advanced research. His main focus is to collaborate with external researchers for research on telecommunication networks. His current network research is focused is on metro Ethernet architecture, optical technologies, Cinegrid, on demand networking, all optical networking. Eric as over 10 year experience of industry and academic research. He graduated with an Honors Bachelor of Electrical Engineering in 1997 and a M.S. in Electrical Engineering in 2000 - photonics telecommunication both degrees from McGill University, Montreal, Canada.

    Li Xu is a Ph.D. student working on the StarPlane project in the SNE (System and Network Engineering) research group at the University of Amsterdam. His current research interests are optical networking, dynamic lightpath provisioning and lightpath authorization. He received his Master degree from the University Leiden in 2006. In 2003 Li Xu graduated with honored Bachelor in computer software engineering from Tsinghua University, Beijing, China and in 2001 he received a B.Sc. in management information systems from Fudan University, Shanghai, China.

    Cees de Laat is associate professor and group leader of the System and Network Engineering Science group in the Informatics Institute at the University of Amsterdam. Research in his group includes optical/switched networking for Internet transport of massive amounts of data in TeraScale eScience applications, RDF to describe networks and associated resources, distributed cross organization Authorization architectures and Systems Security. With SURFnet he develops and implements projects in the GigaPort Research on Networks. He collaborates in the NSF–OptIPuter project. He served in the Open Grid Forum as IETF Liaison and co-chair of the Grid High Performance Networking Research Group (GHPN-RG). He is co-founder and organizer of several of the past meetings of the Global Lambda Integrated Facility (GLIF) and founding member of CineGrid.org. http://www.science.uva.nl/~delaat.

    View full text