Enabling contribution awareness in an overlay broadcasting system

Abstract

We consider the design of bandwidth-demanding broadcasting applications using overlays in environments characterized by hosts with limited and asymmetric bandwidth, and significant heterogeneity in upload bandwidth. Such environments are critical to consider to extend the applicability of overlay multicast to mainstream Internet environments where insufficient bandwidth exists to support all hosts, but have not received adequate attention from the research community. We leverage the multitree framework and design heuristics to enable it to consider host contribution and operate in bandwidth-scarce environments. Our extensions seek to simultaneously achieve good utilization of system resources, performance to hosts commensurate to their contributions, and consistent performance. We have implemented the system and conducted an Internet evaluation on PlanetLab using real traces from previous operational deployments of an overlay broadcasting system. Our results indicate for these traces, our heuristics can improve the performance of high contributors by 10-240% and facilitate equitable bandwidth distribution among hosts with similar contributions.

Stability and performance of overlay multicast systems employing forward error correction

“…In our model the nodes balance their forwarding capacity between the s trees, i.e., a node can have up to ⌈d r /s⌉ children in each of the s trees. One gets the minimum breadth trees described in [3] for s = τ, and the minimum depth trees evaluated in [2,3,11] for s = 1. The case 1 < s < τ was proposed in [16] to improve the overlay's stability under churn.…”

“…The reconnection time consists of the time needed for the detection of the loss of the parent node, the time needed for searching for a new eligible parent node, and the time needed for connecting to the eligible parent. The expected value E[Ξ r e ] of the reconnection time can be up to tens of seconds depending on the tree management and the forwarding capacity in the tree [11].…”

“…We start the evaluation with the minimum depth overlay, that is s = 1, as this is the most common multi-tree-based overlay structure in the literature [2,3,5,11,29,30]. We begin with a homogeneous overlay, and in the following subsections we show how heterogeneity influences the overlay's performance.…”

“…We consider prioritization based on the output capacities of the nodes. A practical alternative would be to consider the number of children of a node [11], as that is easier to estimate, but it would not help high contributor nodes joining the overlay for the first time. This ratio of high and low contributors is similar to that considered in [11] based on a measured trace.…”

“…We consider overlay multicast systems based on multiple distribution trees and the push model, such as the ones in [2][3][4][5][6][10][11][12]. Multiple trees offer two advantages: they ensure graceful quality degradation in dynamic overlays, in which peers can leave during the streaming session and they enable nodes to contribute to the overlay with fractions of the stream bandwidth.…”