Transparent Bandwidth Conservation

The thrust of the work in transparent bandwidth conservation is to examine practical network efficiency, i.e. how can one deliver an efficient network with zero or minimal modifications to the end hosts themselves? Technologies such as multicast have received considerable attention with limited traction in terms of actual deployment. Quite simply, while multicast offers tremendous promise, there exists a significant chasm between a true `native IP multicast' and the curent unicast-centric world. Although ALM (Application Layer Multicast) and AMT (Automatic Multicast Tunneling) offer a better bridge to realizing multicast, the cost of embracing such technology is still significant. Moreover, the fundamental motivations for embracing bandwidth efficiency (i.e. why multicast fared poorly) are still not addressed with current efforts. When the drawbacks of the existing approaches are coupled with the perceived over-abundance of network capacity, the net result is a stalemate with neither content provider nor service provider moving in a significant way towards improving efficiency.

Hence, this research purports to address those fundamental issues to get over that initial roadblock by developing techniques that are legacy application-friendly with directable economic benefit. Critically, we draw on the properties of caching which has been regarded as widely succcessful. In contrast to multicast, web caching and the more recent content distribution networks (CDNs) operate in a manner friendly to the application, i.e. proxying, DNS trickery, etc. that do not require modifications to the applications themselves. Moreover, the economic benefit of such techniques are easily directable and realizable in the current infrastructure without dramatic changes with regards to capability or capacity.

Our work seeks to embrace these key pillars in the form of (1) external transparency and (2) negligible negative QoS impact. By remaining externally transparent, the nearly insurmountable issues of global deployment are avoided by reducing the deployment to a domain-wise or provider-wise. External transparency removes the chicken vs. egg traditionally present in group-wise application deployment (deploy if sufficient demand, develop if deployed, etc., etc., etc.) and directly offers conservation to legacy applications. Moreover, the conservation occurs in an economically directable manner that is easily assessed incentives for the service providers regardless of end host modifications. In order to deliver the external transparency in a legacy friendly manner, our work trades quality of service (QoS) in a limited window, typically enqueuing packets for a brief period of time (0.5-5 ms) to detect redundancy. Our various experimental and simulation studies have noted that we are able to typically achieve a net positive or break even cost in terms of QoS.

This work is funded through the CAREER grant of Dr. Striegel (NSF CNS03-47392). Specific notable outputs of this work include stealth multicast (aka multicast and you didn't even know it), opportunistic wireless broadcast (single hop wireless), and enhanced packet caching (TCP-friendly bandwidth conservation). We believe when combined with an intelligent TCP stack that this work has the potential to revolutionize large scale content distribution with additional benefits for resilience and Denial of Service (DoS) attacks.

Our most recent effort is to encapsulate this work into what we term Scalability in a Box (ScaleBox). We are currently in the process of combining our software to function on COTS hardware as a module in the Click routing software. Advance versions of the code are available upon request for evaluation.

Efforts in this Area

• University Network Tap Analysis
• Stealth Multicast
• PALM - Passive Application Layer Multicast
• EEOD - Explicit End of Data for TCP
• OWB - Opportunistic Wireless Broadcast?
• Tail-Synchronized Media Distribution
• Position Paper - Scalability and Centralization Can Co-Exist
• ScaleBox - Network Scalability in a Box

In addition to our primary efforts, several side efforts have spawned as a result of this CAREER work. The works include:

• Rogue Wireless Access Point Detection - Detecting RWAPs by over-slicing TCP packets at the payload level
• JumboGen? - Improving router scalability by bringing multiple packets together for dynamic jumbo frame transmission with transparent encapsulation/de-capsulation at the ingress/egress

Publications

• D. Salyers, Y. Jiang, A. Striegel, C. Poellabauer, "JumboGen: Dynamic Jumbo Frame Generation for Network Performance Scalability," ACM Computer Communications Review (CCR), vol. 37, no. 5, pp. 53-64, October 2007. CCR Online
• Chad D. Mano, David C. Salyers, Qi Liao, Andrew Blaich, Aaron Striegel, "SAABCOT: Secure Application-Agnostic Bandwidth COnservation Techniques", in Proc. of IEEE BroadNets, pp. 544-550, Raleigh, NC, September 2007 (Invited Paper). DOI IEEE Xplore
• X. Li, A. Striegel, "A Case for Passive Application Layer Multicast," Computer Networks, vol. 51, no. 11, pp. 3157-3171, Aug 2007. DOI BibTeX
• Proposal D. Salyers, "Improving Network Efficiency," Ph.D. Proposal, Adviser: Dr. Aaron Striegel, Committee: Dr. Douglas Thain, Dr. Christian Poellabauer, Dr. Surendar Chandra, March 2007.
• X. Li, D. Salyers, A. Striegel, "Improving Packet Cache Scalability Through the Concept of an Explicit End of Data Marker," in Proc. of HotWeb 2006, Boston, MA, Nov 2006. IEEE Xplore Presentation DOI
• Thesis D. Salyers, "Stealth Multicast: A New Paradigm in Bandwidth Conservation" Masters Thesis, University of Notre Dame, Adviser: Dr. Aaron Striegel, Committee: Dr. Surendar Chandra, Dr. Christian Poellabauer, Dec. 2005.
• D. Salyers, A. Striegel, "A Novel Approach for Transparent Bandwidth Conservation," in Proc. of IFIP Networking, Waterloo, Canada, May 2005.
• C. Mano, A. Striegel, "Trusted Security Devices for Bandwidth Conservation in IPsec Environments," in Proc. of IFIP Networking, Waterloo, Canada, May 2005.