COPS: Quality of Service vs. Any Service at All

  • Randy Katz
  • George Porter
  • Scott Shenker
  • Ion Stoica
  • Mel Tsai
Conference paper

DOI: 10.1007/11499169_1

Part of the Lecture Notes in Computer Science book series (LNCS, volume 3552)
Cite this paper as:
Katz R., Porter G., Shenker S., Stoica I., Tsai M. (2005) COPS: Quality of Service vs. Any Service at All. In: de Meer H., Bhatti N. (eds) Quality of Service – IWQoS 2005. IWQoS 2005. Lecture Notes in Computer Science, vol 3552. Springer, Berlin, Heidelberg

Abstract

Todays networks are awash in illegitimate traffic: port scans, propagating worms, and illegal peer-to-peer transfers of materials [8]. This “noise” has created such a crescendo that legitimate traffic is starved for network resources. Essential network services, like DNS and remote file systems, are rendered unavailable. The challenge is no longer “quality of service” but rather “any service at all”. Techniques must be developed to identify and segregate traffic into good, bad, and suspicious classes. Quality of Service should now protect the good, block the bad, and slow the ugly when the network is under stress of high resource utilization. We discuss the research challenges and outline a possible architectural approach: COPS (Checking, Observing, and Protecting Services). It is founded on “Inspection-and-Action Boxes” (iBoxes) and packet annotations. The former are middlebox network elements able to inspect packets deeply while performing filtering, shaping, and labelling actions upon them. The latter is a new layer between routing and transport that tags packets for control purposes while also providing an in-band control plane for managing iBoxes across a network.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Copyright information

© Springer-Verlag Berlin Heidelberg 2005

Authors and Affiliations

  • Randy Katz
    • 1
  • George Porter
    • 1
  • Scott Shenker
    • 1
  • Ion Stoica
    • 1
  • Mel Tsai
    • 1
  1. 1.CS Division, EECS DepartmentUniversity of CaliforniaBerkeleyUSA

Personalised recommendations