University of Notre Dame NetScale Laboratory

ScaleBox - Transparent Network Capacity Scaling in a Box

ScaleBox

ScaleBox is a culmination of bandwidth conservation techniques developed by Dr. Striegel including enhanced packet caching and stealth multicast coupled with novel techniques for session acceleration and latent synchronization discovery.

  • PC = Packet Caching - Packet payloads are tokenized if the hash of the payload yields a hit in the cache.
  • SM = Stealth Multicast - Packets are enqueued temporarily to aggregate multiple identical payloads into a single multicast packet. The packet is then converted back from multicast before reaching the client.
  • PF = TCP Pre-Fetch - As bandwidth conservation techniques consume much less than their fair bandwidth share on the network, pre-fetching allows the flow to take advantage of that bandwidth slack. In short, hits on the cache are accelerated to enable a fast pre-fetch when future tokenized packets are highly likely. The ramp up time is then based on the local round trip time (LRTT) rather than the actual RTT, offering a dramatically accelerated content grab while still taking significantly less bandwidth on the bottleneck links.

Item Description
Papers Grab papers related to ScaleBox.
Presentations Grab presentations related to ScaleBox.
Code Download the code for ScaleBox from this website via Subversion or a tar/gzip file. The code is currently in alpha development with builds supported under most UNIX versions as well as the NGW 100 embedded platform.
Manual Examine the manual for the ScaleBox code.
Support Support for this work was provided in part by the National Science Foundation, Intel, and Sun Microsystems.
r5 - 23 Jun 2008 - 20:08:58 - AaronStriegel
This site is powered by the TWiki collaboration platformCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback
Syndicate this site RSSATOM