English
 
User Manual Privacy Policy Disclaimer Contact us
  Advanced SearchBrowse

Item

ITEM ACTIONSEXPORT

Released

Conference Paper

SAAR: A Shared Control Plane for Overlay Multicast

MPS-Authors
/persons/resource/persons180339

Nandi,  Animesh
Group P. Druschel, Max Planck Institute for Software Systems, Max Planck Society;

/persons/resource/persons144511

Druschel,  Peter
Group P. Druschel, Max Planck Institute for Software Systems, Max Planck Society;

External Ressource
No external resources are shared
Fulltext (public)
There are no public fulltexts stored in PuRe
Supplementary Material (public)
There is no public supplementary material available
Citation

Nandi, A., Ganjam, A., Druschel, P., Ng, T., Stoica, I., Zhang, H., et al. (2007). SAAR: A Shared Control Plane for Overlay Multicast. In Proceedings of the 4th USENIX Symposium on Networked Systems Design & Implementation (NSDI'07) (pp. 57-72). Berkeley, USA: USENIX.


Cite as: http://hdl.handle.net/11858/00-001M-0000-0028-8C78-4
Abstract
Many cooperative overlay multicast systems of diverse designs have been implemented and deployed. In this paper, we explore a new architecture for overlay multicast: we factor out the control plane into a separate overlay that provides a single primitive: a configurable anycast for peer selection. This separation of control and data overlays has several advantages. Data overlays can be optimized for efficient content delivery, while the control overlay can be optimized for flexible and efficient peer selection. Several data channels can share a control plane for fast switching among content channels, which is particularly important for IPTV. And, the control overlay can be reused in multicast systems with different data plane organizations. We designed and evaluated a decentralized control overlay for endsystem multicast. The overlay proactively aggregates system state and implements a powerful anycast primitive for peer selection. We demonstrate that SAAR“s efficiency in locating peers reduces channel switching time, improves the quality of content delivery, and reduces overhead, even under dynamic conditions and at scale. An experimental evaluation demonstrates that the system can efficiently support single-tree, multitree and block-based multicast systems.