Information-driven network resilience: Research challenges and perspectives

https://doi.org/10.1016/j.osn.2016.06.002Get rights and content

Abstract

Internet designed over 40 years ago was originally focused on host-to-host message delivery in a best-effort manner. However, introduction of new applications over the years have brought about new requirements related with throughput, scalability, mobility, security, connectivity, and availability among others. Additionally, convergence of telecommunications, media, and information technology was responsible for transformation of the Internet into an integrated system enabling accessing, distributing, processing, storing, and managing the payload of these messages.

Users are now visibly more interested in receiving / accessing information independently of the network location of its host. This consideration in turn revived the interest in named data-driven networking (a.k.a. Information-Centric Networking - ICN). Instead of assuming that networks are limited to the manipulation of network locator space, the basic assumption underneath is that information can be named, addressed, and matched independently of its network location leaving in turn the possibility to match message delivery delay requirements.

In this paper, we summarize our research conducted in order to bring a completely different view / perspective of network resilience, originally defined as the ability of a network to assure an acceptable level of service in the face of various faults and challenges to normal operation. That is, instead of maintaining network reachability independently of its actual utility to the “end-points”, our research aimed at exchanging and confronting the key principles that would enable an information-driven resilience (networked) scheme. More precisely, knowing that the user utility function is mainly driven nowadays by information-related criteria such as accessibility (reachability), how to design network resilience schemes that would be directed toward that goal. The main challenge is thus: can one design resilience schemes that combine maximization of end-point utility function and minimization of the network-related cost?

Introduction

Following the seminal work of Baran [1] in the early 60's, information is segmented into packets before being transmitted towards the destination (identified by its network attachment point a.k.a. network locator). Without anticipating the main consequences of such decomposition, Baran laid actually the distinction between names and addresses. Moreover as a result of this decomposition (between application and network data units), functionality of the network layer has been since more than 30 years confined to destination-based packet forwarding along logical communication channels identified by their (destination) address/network locator (connectivity function). Users are now visibly more interested in receiving/accessing information independently wherever it is located rather than accessing a particular node (or a host/server). This in turn gives rise to the Information-Centric Networking (ICN) concept [2], [3] with the basic assumption that information can be named, addressed, and matched independently of its network physical location leaving in turn the possibility to match message delivery delay requirements. However, nowadays networks are mainly used for information exchanges (distribution function).

On the other hand, in response to the increasing Internet traffic volume – 50% per year following the Nielsen Law [4] – for applications such as (mobile) video-streaming, social networking and cloud computing, various proprietary technologies enabling content distribution have been developed as overlays that rely on caching and replication. The recent advent of Content Service Providers “in between” Application Service Provider (ASP) and Internet Service Provider (ISP) is a representative of this evolution. However, being mostly deployed in silos (a silo can be defined by the application-specific information space delimited by its own naming and exchange conventions), it becomes impossible to uniquely and securely identify named information independently of the communication channel. Moreover, as these different content distribution technologies are typically implemented as an overlay, they lead to unnecessary inefficiency in particular, in terms of resource consumption and mobilization.

The recent Information-Centric Networking (ICN) out of the seminal work initiated in the 70's aims at enabling data to become independent from their network location, application, and storage support [2], feasible by means of content exchanges enabling in turn in-network caching and replication. As information/digital objects can be duplicated and made available/segmented at multiple and heterogeneous storage entities/locations, decoupling the information object from its network location would lead to support of mobility of both information and hosts. At the end of the spectrum:

  • Internet Protocol (IP) destination addresses have a local scope,

  • associated forwarding decisions are limited to link-local exchanges, and

  • information exchange / relationship replaces end-to-end connectivity established through packet-oriented communication channels.

Thus information-centric networking questions more the design of Transmission Control Protocol (TCP) and its emulation of communication channels associated to the data (precisely, the segmentation of Internet Messaging Protocol (IMP) into TCP and IP) rather than “packetization” provided by IP.

Remember that IP locators are used as part of the transport address. All recent Hypertext Transfer Protocol (HTTP) developments in incorporating flow control (but also real-time adaptive behaviour) can be considered as the application-layer answer to this architectural flaw.

Many challenges remain of course to be addressed along this evolution, in particular, with respect to the forwarding plane and its scaling properties. Indeed, name spaces have not been designed to sustain forwarding performance and forwarders scaling contrary to IP addresses, which can be aggregated, summarized, and translated, as well as subject to a decentralized control.

It also important to emphasize that the problem space does not limit to the common network optimization problem of minimizing resource consumption or maximizing utility. It involves functional transformation that was not technologically possible at the first stage of the “Internet” whose communication stack has been designed such as to minimize packet losses in case of disruptions. Packetization, by decoupling communications from the physical channel, thus aims at providing a primary form of resilience (defined e.g., in [5], [6] as the ability of a network to assure an acceptable level of service in the face of various faults and challenges to normal operation). In this respect, information networking opens a much wider and fundamental issue: beyond circuit and packet switching, which forwarding paradigm fits best information exchanges?

In this paper, we focus on the research challenges related to the resilience properties a global information distribution system would have to meet. Indeed, these evolutions also lead to major rethinking of traditional objectives such as “availability”, since moving from destination reachability and its association to physical topology/path to information accessibility and availability leads to completely different challenges. Indeed, the same/similar information may be (and is often) available at different locations subject to various resilience capabilities. Hence, resilience is not limited anymore to communication functionality of the network but also couples its storage and even processing capability.

Overall, we could state that since so far, most approaches to communication networks resilience were designed as if the requested data was located at a very limited number of locations. These locations sit at the periphery of the network and involve establishment of an end-to-end communication channel before reaching it (implying in turn that all intermediate nodes would be configured to reach the corresponding destination). Hence, packet and circuit-oriented networks share similar properties in that respect as they are both centred on the key principle that the underlying layer provides the mechanism for the upper-layer to perform following its expectations.

Information-driven networks consider that exchange patterns are instead driven from the top and that the functionality of the network is to provide a distributed search process and corresponding forwarding decisions that would best cope with the user utility. Indeed, maintaining a forwarding state, a fully qualified domain name (FQDN) prefix would increase the number of forwarding entries in transit nodes from the level of 106 to 1015–1018.

It is important to remember here that there is no trivial aggregation or summarization of object names (and their attributes) compared to IP addressing beyond hierarchically structured naming. The numbers of entries become even larger if one considers “objects” beyond files. This implies that most procedures and algorithms would have to be supra-linear (i.e., less than linear in the input space size) and ideally logarithmic in both time and resource complexity, or even better scale-free [7]. In turn, applying this principle brings back the user utility function in the loop instead of a quasi-permanent delegation from the host to the network with separate concerns.

Admittedly, this challenge has never been adequately addressed by the Internet architecture, which is elaborated around a distinction between host / terminal and network performing non-cooperatively. The latter is transparent to the former besides requiring the address of the exit gateway. In this respect, information networking may also pave the way to new cooperation modes. Interestingly enough, the relation at the attachment points of the information graph would enable information access over a wider spectrum of physical media.

Alternatively, if the distribution functionality of information networks is conceived and designed following the same principles as those underlying the current forwarding system, then per-flow state maintenance will also remain a first-order design principle and forwarding table scaling a key design goal (many ICN research initiatives have worked on this alternative during the last decade).

In this context, extending the notion of resilience can follow two paths, which also corresponds to the distinction made in this paper between ICN (and its variants) vs. information networks. The former is positioned from the perspective of extending the “network functionality” to better accommodate name-data object/file exchanges compared to content distribution network (CDN) overlays. Hence, the main challenge sits at the algorithmic level keeping scaling challenges as first design principle.

On the other hand, information networks entail a broader perspective than ICN by supporting all information exchanges beyond data object/file asynchronous transactions such as (near-) real-time human-to-human, machine-to-machine but also human-to-machine, and machine-to-human information exchanges. In other terms, information networks preserve the generic purpose of the shared “infrastructure” which is paradoxically broken by augmenting/optimizing some parts of the network for file exchanges. Broadly speaking, the main challenge includes the inception of new processes that would lead to new classes of procedures in particular, to address the resilience challenge.

In the remaining part of the paper, we first review in Section 2.1 the classical communication network availability and resilience concepts, as well as highlight the main issues when combining the information distribution/exchange in the problem space. Section 2.2 in turn outlines the concept underlying the information-driven networking, and relation of ICN with peer-to-peer paradigm. Section 3 addresses the resilience challenges related with ICN concepts, including: Overlay Network model (Section 3.1), Named-data routing model (3.2 Named-data routing model, 3.3 Works and challenges related to ICN resilience for the named-data model and beyond), Peer-to-peer model (Section 3.4), and discusses example case studies (Section 3.5) as well as disaster resilience issues (Section 3.6). The main focus of Section 4 is on practical implementation issues concerning Software-Defined Networking (SDN)/OpenFlow. Section 5 concludes the paper.

Section snippets

Communications resilience

Nowadays, physical infrastructure of Internet is mostly supported by optical communication systems. In the classical resilience approach, the challenge is to design an IP/MPLS network over a DWDM physical layer meeting both traffic and network capacity constraints. The requirements include e.g., resilience against single-point of failures (by means of 2-node connectivity in order not to encounter service disruption), and mapping from logical into physical layer (trying to save bandwidth and

Challenges related to concepts of information-centric networking and resilience

According to Shoch [53], [54], the “name” of a resource indicates what we seek, an “address” indicates where it is, and a “route” tells us how to get there. This distinction led indeed to considering that the routing function performs on addresses (network locators) and the resulting (routing) topology is stable before performing name-to-address resolution for localization purposes. In other words, localization relies on connectivity. Further, these functions operate depending on the spatial

Resilience issues of information delivery via SDN / OpenFlow

The need of transition from conventional networking to information-driven networking schemes has also its root in complexity and rather hard management of IP networks mainly due to common integration of data and control planes (Fig. 14a). In particular, this integration forces network operators to configure each network element separately by using the low-level device-centric (and often vendor-specific) commands. As a result, resilience becomes a major issue mainly due to common network

Conclusions

In this paper, we focused on research challenges and perspectives related with resilience of information-driven networking. This concept was originally proposed as a response to evolution of applications over the years requiring more and more throughput, improved mobility, scalability, security, etc. The convergence of telecommunications, media, and information technology has initiated the transformation of Internet - the global communications infrastructure - into an integrated system with

Acknowledgements

This article is based upon work from COST Action CA15127 (“Resilient communication services protecting end-user applications from disaster-based failures” single bond RECODIS), supported by COST (European Cooperation in Science and Technology).

References (134)

  • A. Avizienis et al.

    Basic concepts and taxonomy of dependable and secure computing

    Trans. Dependable Secur. Comput.

    (2004)
  • P. Rodríguez-Bocca, Quality-centric design of peer-to-peer systems for live-video broadcasting, (Ph.D. Thesis),...
  • E.K. Cetinkaya, J.P.G. Sterbenz, A taxonomy of network challenges, in: Proceedings of the 9th International Conference...
  • J. Cowie, A. Popescu, T. Underwood, Impact of hurricane Katrina on Internet infrastructure Technical Report, Renesys,...
  • Y. Kitamura et al.

    Experience with restoration of Asia Pacific network failures from Taiwan earthquake

    IEICE Trans. Commun.

    (2007)
  • Y. Ran

    Considerations and suggestions on improvement of communication network disaster countermeasures after the Wenchuan earthquake

    IEEE Commun. Mag.

    (2011)
  • S. Urushidani et al.

    Highly available network design and resource management of SINET4

    Telecommun. Syst.

    (2014)
  • J. Kappenman

    A perfect storm of planetary proportions

    IEEE Spectr. Mag.

    (2012)
  • J. Rak

    A new approach to design of weather disruption-tolerant Wireless Mesh Networks

    Telecommun. Syst.

    (2016)
  • J. Rak

    LLA: a new anypath routing scheme providing long path lifetime in VANETs

    IEEE Commun. Lett.

    (2014)
  • J. Rak

    Providing differentiated levels of service availability in VANET communications

    IEEE Commun. Lett.

    (2013)
  • A. Avizienis, J.C. Laprie, B. Randell, Dependability and its threats: a taxonomy, in: R. Jacquart (ed.), Building the...
  • W.D. Grover, Mesh-based Survivable Networks. Options and Strategies for Optical, MPLS, SONET, and ATM Networks,...
  • B. Ramamurthy et al.

    Survivable WDM mesh networks

    IEEE/OSA J. Light. Technol.

    (2003)
  • D. Kutscher, (Ed.) ICN research challenges, Internet Research Task Force (IRTF), Work in progress, February...
  • P. Chołda et al.

    Quality of Resilience as a network reliability characterization tool

    IEEE Netw.

    (2009)
  • P.-H. Ho

    State of the art progress in developing survivable routing schemes in mesh WDM networks

    IEEE Commun. Surv. Tutorials

    (2004)
  • A. Haider et al.

    Recovery techniques in next generation networks

    IEEE Commun. Surv. Tutorials

    (2004)
  • P. Chołda et al.

    Recovery and its quality in multilayer networks

    IEEE/OSA J. Light. Technol.

    (2010)
  • W. Molisz et al.

    Region Protection/restoration Scheme in Survivable Networks, Lecture Notes in Computer Science

    (2005)
  • J. Rak

    Fast service recovery under shared protection in WDM networks

    IEEE/OSA J. Light. Technol.

    (2012)
  • D. Schupke

    Multilayer and multidomain resilience in optical networks

    Proc. IEEE

    (2012)
  • P. Chołda et al.

    A survey of resilience differentiation frameworks in communication networks

    IEEE Commun. Surv. Tutorials

    (2007)
  • D.R. Karger et al.

    A randomized linear-time algorithm to find minimum spanning trees

    J. ACM

    (1995)
  • M.F. Al-Naday et al.

    Information resilience: source recovery in an information-centric network

    IEEE Netw.

    (2014)
  • V.G. Cerf

    The day the Internet age began

    Nature

    (2009)
  • T. Berners-Lee et al.

    World-Wide Web: the information universe

    Electron. Netw.: Res. Appl. Policy

    (1992)
  • G. Xylomenos et al.

    A survey of Information-Centric Networking research

    IEEE Commun. Surv. Tutorials

    (2014)
  • G. Sallai, Chapters of Future Internet research, in: Proceedings 4th International Conference on Cognitive...
  • B. Carlsson, R. Gustavsson, The rise and fall of Napster - an evolutionary approach, in: Proceedings of the 6th...
  • E. Adar et al.

    Free riding on Gnutella

    First Monday

    (2000)
  • M. Ripeanu, Peer-to-peer architecture case study: Gnutella network, in: Proceedings of the 1st International Conference...
  • B. Cohen, Incentives build robustness in BitTorrent, in: Proceedings of the Workshop on Economics of Peer-to-Peer...
  • Z. Ma, D. Qiu, A novel optimistic unchoking algorithm for BitTorrent, in: Proceedings of the 6th IEEE Consumer...
  • R. Axelrod

    The Evolution of Cooperation

    (1984)
  • P. Romero, Mathematical analysis of scheduling policies in peer-to-peer video streaming networks, (Ph.D. dissertation),...
  • J. Pan et al.

    A survey of the research on future Internet architectures

    IEEE Commun. Mag.

    (2011)
  • H.-Z. Chou et al.

    Randomised and distributed methods for reliable peer-to-peer data communication in wireless ad hoc networks

    IET Commun.

    (2007)
  • B. Gedik et al.

    A scalable peer-to-peer architecture for distributed information monitoring applications

    IEEE Trans. Comput.

    (2005)
  • K. Walkowiak et al.

    Simultaneous optimization of unicast and anycast flows and replica location in survivable optical networks

    Telecommun. Syst.

    (2013)
  • Cited by (23)

    • Transport SDN at the dawn of the 5G era

      2019, Optical Switching and Networking
    • Redundancy Cyber Resiliency Technique Based on Fast ReRouting under Security Metric

      2021, 2020 IEEE International Conference on Problems of Infocommunications Science and Technology, PIC S and T 2020 - Proceedings
    • Design and research of the model for secure traffic engineering fast reroute under traffic policing approach

      2021, Experience of Designing and Application of CAD Systems in Microelectronics
    View all citing articles on Scopus
    View full text