Middleware support for multicast-based data dissemination: a working reality

Chrysanthis, Panos K.; Liberatore, Vincenzo; Pruhs, Kirk

doi:10.1109/words.2003.1218092

Cited by 14 publications

(12 citation statements)

References 69 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Its motivation came from a problem related to design of cyclic multicast schedule. Considering current trend in technologies and applications, cyclic multicast that pays heed to data dependencies should play a pivotal role in the future [6]. Our research pointed out certain negative aspects of the MCA problem, namely, it does not have a polynomial time algorithm and it does not even admit a polynomial time approximation scheme for arbitrary graph instance (unless P=NP).…”

Section: Discussionmentioning

confidence: 93%

On Minimum Circular Arrangement

Ganapathy

Lodha²

2004

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Motivated by a scheduling problem encountered in multicast environments, we study a vertex labelling problem, called Minimum Circular Arrangement (MCA), that requires one to find an embedding of a given weighted directed graph into a discrete circle which minimizes the total weighted arc length. Its decision version is already known to be NP-complete when restricted to sparse weighted instances. We prove that the decision version of even un-weighted MCA is NP-complete in case of sparse as well as dense graphs.We also consider complementary version of MCA, called MaxCA. We prove that it is MAX-SNP[π] complete and, therefore, has no PTAS unless P=NP. A similar proof technique shows that MCA is MAX-SNP[π]-Hard and hence admits no PTAS as well. Then we prove a conditional lower bound of √ 2 − for MCA approximation under some hardness assumptions, and conclude with a PTAS for MCA on dense instances.

show abstract

Section: Discussionmentioning

confidence: 93%

On Minimum Circular Arrangement

Ganapathy

Lodha²

2004

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…We first present the configuration of our system, which is shown in Figure 1, for more information on this architecture please see [14,26]. The configuration demonstrates the main building blocks, each tied to a particular data management function in multicast environments (document selection, scheduling, consistency, caching, and indexing).…”

Section: Our Hybrid Systemmentioning

confidence: 99%

“…Additionally, if multicast push were to be used extensively, the multicast channel could be clogged with unpopular items, leading to increased overall response times for clients. Therefore, a more prudent approach is to have multicast push be reserved for popular documents, and less popular data can be disseminated with the traditional request-response unicast pull approach [2,14,31]. This we refer to as a hybrid system approach to data dissemination because of its combination of multiple delivery types to handle the distribution of data.…”

Section: Introductionmentioning

confidence: 99%

Improving the Hybrid Data Dissemination Model of Web Documents

et al. 2008

View full text Add to dashboard Cite

One of the major problems in the Internet today is the scalable delivery of data. With more and more people joining the Internet community, web servers and services are being forced to deal with workloads beyond their original data dissemination design capacity. One solution that has arisen to address scalability is to use multicasting, or push-based data dissemination, to send out data to many clients at once. More recently, the idea of using multicasting as part of a hybrid system with unicasting has shown positive results in increasing server scalability. In this paper we focus on solving problems associated with the hybrid dissemination model. In particular, we address the issues of document popularity and document division while arguing for the use of a third channel, called the multicast pull channel, in the hybrid system model. This channel improves performance in terms of response time while improving the robustness of the hybrid system. We show through extensive simulation using our working hybrid server the usefulness of this additional channel and its improving effects in creating a more scalable and more efficient web server.

show abstract

“…Multicast may also arise in a wired network as a method to provide scalable data dissemination. See for example multicast based scalable data dissemination systems described in [6,7]. One commercial example of a multicast-pull client-server system is Hughes' DirecPC system [8].…”

Section: Introductionmentioning

confidence: 99%

A maiden analysis of longest wait first

Edmonds

Pruhs

2005

ACM Trans. Algorithms

View full text Add to dashboard Cite

We consider server scheduling strategies to minimize average flow time in a multicast pull system where data items have uniform size. The algorithm Longest Wait First (LWF) always services the page where the aggregate waiting times of the outstanding requests for that page is maximized. We provide the first non-trivial analysis of the worst case performance of LWF. On the negative side, we show that LWF is not s-speed O(1)-competitive for s <. On the positive side, we show that LWF is 6-speed O(1)-competitive.

show abstract

Middleware support for multicast-based data dissemination: a working reality

Cited by 14 publications

References 69 publications

On Minimum Circular Arrangement

On Minimum Circular Arrangement

Improving the Hybrid Data Dissemination Model of Web Documents

A maiden analysis of longest wait first

Contact Info

Product

Resources

About