Presentation on theme: "Virtual Distance: A Generalized Metric for Overlay Tree Construction ISCC12 July 1 2012 Suat Mercan (Zirve University) & Murat Yuksel (University of Nevada,"— Presentation transcript:
Virtual Distance: A Generalized Metric for Overlay Tree Construction ISCC12 July 1 2012 Suat Mercan (Zirve University) & Murat Yuksel (University of Nevada, Reno)
Motivation Increasing bandwidth (127 Kbps(2000) 4.4 Mbps(2010)) Increasing Internet usage 360,985,492(2000) - 1,966,514,816 (2010) High demand for multimedia applications (Internet TV, tele-conference, online education, youtube) A huge amount of internet traffic coming from multimedia (90% of Internet traffic in 2014 according to Cisco) This trend is calling efficient and scalable mechanisms to transfer content to many receivers from a single source
Motivation Various types of multimedia applications Sensitivity of these applications differs against various network performance metrics – delay is crucial for video conferencing which includes interactivity – jitter is important for video streaming – bandwidth is for HDTV
Our Work Responding to these varying application-specific requirements or sensitivities is a key point We define and use the concept of virtual distance for constructing overlay trees We aim to find the most appropriate parent for a peer according to the applications purpose Calculating the virtual distance based on different criteria, but without protocol modification, makes the overlay multicast protocol satisfy different quality expectations for different apps Our key goal is to automatically calculate overlay multicast trees such that they can be seamlessly customized to applications performance goals.
Virtual Direction Multicast We embed the virtual distance method in our previous overlay multicast tree protocol, VDM. VDM: – Tree based – Source is alone at the beginning – Tree extends with new joins. – Want to use minimum resource as satisfying end user
Round Trip Time (RTT) is used to calculate distances What about loss rate? Delay and loss rate between two nodes may be uncorrelated because of background and cross traffic on routers A peer might experience high loss rate on a good path in terms of delay Sensitivity of multimedia applications differs against various network performance metrics This requires to take other factors into account when building overlay tree
Virtual Distance A key property of VDM is the capability of virtualizing the underlying network in different ways It is possible to establish virtual directions based on performance metrics Different values of these metrics may produce different virtual distances and thus different overlay tree We can establish target specific overlay trees to improve some specific performance metrics
Information from iPlane Dataset Inter-PoP measurement data sample from iPlane Three points A, B, C d1 = distance (A-B) d2 = distance (B-C) l1 = loss(A-B) l2 = loss(B-C) d1/d2 and l1/l2 are inversely correlated for 44%
We evaluate the performance of VDM-D (delay-based) and VDM-L (loss-based) in order to show the efficiency of the virtual distance concept We analyze protocol behaviors as we vary the churn rate in the overlay network
Simulation Setup NS-2 (Network simulator) 792 nodes in physical network 200 nodes in overlay network Degree of each node between 2-5 Each physical link is assigned error rate between 0% - 2% Run 5,000s Under different churn rates (1% - 20%) Run each simulation 10 times
Simulation Results Stretch Loss VDM optimizes the loss when the virtual distance is based on loss. VDM optimizes the stretch when the virtual distance is based on delay.
Conclusions Our key goal is to automatically calculate overlay multicast trees such that they can be seamlessly customized to applications performance goals. VDM-D uses delay for distance estimation, and improves stress and stretch while giving higher loss rate VDM-L achieves better performance in terms of loss rate