On the treeness of internet latency and bandwidth

Ramasubramanian, Venugopalan; Malkhi, Dahlia; Kühn, Fabian; Balakrishnan, Mahesh; Gupta, Archit; Akella, Aditya

doi:10.1145/1555349.1555357

Cited by 58 publications

(86 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…First, Ramasubramanian et. al [58] verify that a bandwidth data set produced many small ε values. ε was introduced by Abraham et.…”

Section: Network Bandwidth In a Tree Metric Spacementioning

confidence: 86%

Decentralized Network Bandwidth Prediction and Node Search

Song

2011

2011 IEEE International Symposium on Parallel and Distributed Processing Workshops and PHD Forum

View full text Add to dashboard Cite

As modern computing becomes increasingly data-intensive and distributed, it is becoming crucial to effectively manage and exploit end-to-end network bandwidth information from hosts on wide-area networks. Inspired by the finding that Internet bandwidth can be represented approximately in a tree metric space, we focus on three specific research problems. First, we have designed a decentralized algorithm for network bandwidth prediction. The algorithm embeds the bandwidth information as distance in an edgeweighted tree, without performing full n-to-n measurements. No central and fixed infrastructure is required. Each joining node performs a limited number of sampling measurements. Second, we designed a decentralized algorithm to search for a centroid node that has high-bandwidth connections with a given set of nodes.The algorithm can find a centroid accurately and efficiently using the bandwidth data produced by the prediction algorithm. Last, we have designed another type of decentralized search algorithm to find a cluster of nodes that have high-bandwidth interconnections. While the clustering problem is NP-complete in a general graph, our algorithm runs in polynomial time with the bandwidth data predicted in a tree metric space. We provide proofs that our algorithms for bandwidth prediction and node search have perfect accuracy and high scalability when a network is modeled as a tree metric space. Also, experimental results with real-world data sets validate the high accuracy and scalability of our approaches.

show abstract

“…First, Ramasubramanian et. al [58] verify that a bandwidth data set produced many small ε values. ε was introduced by Abraham et.…”

Section: Network Bandwidth In a Tree Metric Spacementioning

confidence: 86%

Decentralized Network Bandwidth Prediction and Node Search

Song

2011

2011 IEEE International Symposium on Parallel and Distributed Processing Workshops and PHD Forum

View full text Add to dashboard Cite

show abstract

“…A natural question is whether we could provide more structure to the communication cost function by reverse engineering this underlying network topology. Approaches such as Sequoia [53] deduce underlying network topology by mapping application nodes onto leaves of a virtual tree. Unfortunately, even though these inference approaches work well for Internet-level topologies, state-of-the-art methods cannot infer public cloud environments accurately [10,16].…”

Section: Cost Functionsmentioning

confidence: 99%

ClouDiA: a deployment advisor for public clouds

et al. 2015

View full text Add to dashboard Cite

An increasing number of distributed data-driven applications are moving into shared public clouds. By sharing resources and operating at scale, public clouds promise higher utilization and lower costs than private clusters. To achieve high utilization, however, cloud providers inevitably allocate virtual machine instances non-contiguously; i.e., instances of a given application may end-up in physically distant machines in the cloud. This allocation strategy can lead to large differences in average latency between instances. For a large class of applications, this difference can result in significant performance degradation, unless care is taken in how application components are mapped to instances. In this paper, we propose ClouDiA, a general deployment advisor that selects application node deployments minimizing either (i) the largest latency between application nodes, or (ii) the longest critical path among all application nodes. ClouDiA employs a number of algorithmic techniques, including mixed-integer programming and constraint programming techniques, to efficiently search the space of possible mappings of application nodes to instances. Through experiments with synthetic and real applications in Amazon EC2, we show that mean latency is a robust metric to model

show abstract

“…Another special case is when only one red node is left. In that case (see lines [9][10][11][12][13][14], the algorithm chooses at each step the node with the largest b i (unless it is red and G(π) < T ).…”

Section: B Greedy Algorithmmentioning

confidence: 99%

“…to give them coordinates) so that their distance in the metric space estimates well the metric of interest (usually the latency between two nodes or the bandwidth a communication between them can achieve). In the case of latencies, the most well known embedding tools are Vivaldi [12], which embeds nodes into a 2D+1 metric space and relies on direct measurements to adapt dynamically node coordinates, and Sequoia [13], which embeds the nodes as the leaves of a weighted tree and relies on the distance in the tree to estimate the latency. Both coordinate systems have been extended to estimate bandwidths in [13], but it has been recently proved experimentally [14] on the PlanetLab dataset that the basic LastMile or bounded multiport model, where each node is associated to a incoming and an outgoing bandwidth limit, and where the achievable bandwidth between C i and C j is the minimum of the outgoing bandwidth of C i and the incoming bandwidth of C j , is more accurate than those more sophisticated models with respect to bandwidth prediction.…”

Section: Introductionmentioning

confidence: 99%

Broadcasting on large scale heterogeneous platforms under the bounded multi-port model

Beaumont

Eyraud-Dubois

Agrawal

2010

2010 IEEE International Symposium on Parallel &Amp; Distributed Processing (IPDPS)

View full text Add to dashboard Cite

Abstract-We consider the classical problem of broadcasting a large message at an optimal rate in a large scale distributed network. The main novelty of our approach is that we consider that the set of participating nodes can be split into two parts: "green" nodes that stay in the open-Internet and "red" nodes that lie behind firewalls or NATs. Two red nodes cannot communicate directly, but rather need to use a green node as a gateway for transmitting a message. In this context, we are interested in both maximizing the throughput (i.e. the rate at which nodes receive the message) and minimizing the degree at the participating nodes, i.e. the number of TCP connections they must handle simultaneously. We both consider cyclic and acyclic solutions for the flow graph. In the cyclic case, our main contributions are a closed form formula for the optimal cyclic throughput and the proof that the optimal solution may require arbitrarily large degrees. In the acyclic case, we prove that it is possible to achieve the optimal throughput with low degree. Then, we prove a worst case ratio between the optimal acyclic and cyclic throughput and show through simulations that this ratio is on average very close to 1, which makes acyclic solutions efficient both in terms of the throughput and the number of connections.

show abstract

On the treeness of internet latency and bandwidth

Cited by 58 publications

References 41 publications

Decentralized Network Bandwidth Prediction and Node Search

Decentralized Network Bandwidth Prediction and Node Search

ClouDiA: a deployment advisor for public clouds

Broadcasting on large scale heterogeneous platforms under the bounded multi-port model

Contact Info

Product

Resources

About