Citation Context

... been observed that some of these overlays exhibit properties of small-world and scale-free networks [1, 11]. However, in many cases it is better to construct overlays that are close to random graphs =-=[3]-=-. Gossip-based unstructured overlays can be used as a building block in a variety of network management applications, especially when highly dynamic environments are anticipated. A good example is sel...

Citation Context

...rent gossiping protocol, Newscast [7, 16], in managing P2P routing tables. In particular, we exploit Newscast’s randomized overlay to pick nodes that satisfy certain criteria, to fill in Pastry-like (=-=[13]-=-) routing tables. CYCLON is, in fact, a better candidate than Newscast in that application, as it exhibits a superior randomness property and lower bandwidth requirements 14scompared to Newscast. Fina...

Citation Context

...tworks generally exhibit self-healing behavior with respect to major network disasters. It has also been observed that some of these overlays exhibit properties of small-world and scale-free networks =-=[1, 11]-=-. However, in many cases it is better to construct overlays that are close to random graphs [3]. Gossip-based unstructured overlays can be used as a building block in a variety of network management a...

Citation Context

...ver, as is shown in [6], there is a large family of membership protocols for unstructured overlays for which this assumption is false. This includes the work on lightweight probabilistic multicasting =-=[4]-=-, as well as the Newscast system [7, 16]. As it turns out, such membership protocols generally lead to small-world graphs, which distinguish themselves from random graphs by a high clustering coeffici...

Citation Context

...vu.nl 1sThere are currently two main categories of P2P systems in terms of overlay management. Structured P2P systems impose a specific linkage structure between nodes. Distributed Hash Tables (DHTs) =-=[2]-=- are typical examples in this category. They link peers based on their IDs in a way to enable efficient ID-based routing among them. In contrast, in unstructured P2P systems peers are not linked accor...

Citation Context

...roperties with random graphs. We have already described basic shuffling, introduced in [14], forming the starting point for our own work described in this paper. Another example is the Scamp protocol =-=[5]-=-. Scamp is reactive, in the sense that cache exchanges take place only when nodes join, leave, or a failure is detected. As it turns out, the protocol exhibits similar properties in comparison to rand...

Citation Context

...arily designed to support rapid information dissemination and content-based searching in highly dynamic distributed environments. In this paper we concentrate on gossip-based unstructured P2P systems =-=[6]-=-. These systems aim at exploiting randomness to disseminate information across a large set of nodes. A key issue is to keep the overlay connected even in the event of major disasters without maintaini...

Citation Context

...So far, shuffling appears to fall outside the category of small-world networks. Of course, random graphs may not be the best structure for communication networks. Alternative schemes are described in =-=[12, 10]-=-. In these cases, the requirements of low diameter, resilience to massive node failures, and inexpensive membership management lead to specific graph-construction protocols. We argue that with enhance...

Citation Context

...our prior work on managing routing tables in very large P2P networks [17]. Stavrou et al. suggest management of flash crowd crises by disseminating information over gossip-based unstructured overlays =-=[14]-=-. In [18] we build on top of the research presented in this paper to maintain links between semantically-related peers, and we show that it substantially enhances distributed searching for content (i....

Citation Context

...So far, shuffling appears to fall outside the category of small-world networks. Of course, random graphs may not be the best structure for communication networks. Alternative schemes are described in =-=[12, 10]-=-. In these cases, the requirements of low diameter, resilience to massive node failures, and inexpensive membership management lead to specific graph-construction protocols. We argue that with enhance...

Citation Context

...tworks generally exhibit self-healing behavior with respect to major network disasters. It has also been observed that some of these overlays exhibit properties of small-world and scale-free networks =-=[1, 11]-=-. However, in many cases it is better to construct overlays that are close to random graphs [3]. Gossip-based unstructured overlays can be used as a building block in a variety of network management a...

Citation Context

...ate the document, the client node performs a series of randomized, scoped searches over the links of the overlay created by basic shuffling. In [17], we apply a different gossiping protocol, Newscast =-=[7, 16]-=-, in managing P2P routing tables. In particular, we exploit Newscast’s randomized overlay to pick nodes that satisfy certain criteria, to fill in Pastry-like ([13]) routing tables. CYCLON is, in fact,...

Citation Context

... work on managing routing tables in very large P2P networks [17]. Stavrou et al. suggest management of flash crowd crises by disseminating information over gossip-based unstructured overlays [14]. In =-=[18]-=- we build on top of the research presented in this paper to maintain links between semantically-related peers, and we show that it substantially enhances distributed searching for content (i.e., in fi...

Citation Context

...ate the document, the client node performs a series of randomized, scoped searches over the links of the overlay created by basic shuffling. In [17], we apply a different gossiping protocol, Newscast =-=[7, 16]-=-, in managing P2P routing tables. In particular, we exploit Newscast’s randomized overlay to pick nodes that satisfy certain criteria, to fill in Pastry-like ([13]) routing tables. CYCLON is, in fact,...

Citation Context

... load fairness, and can guarantee that no node is left unattended, resulting in a robust selfmonitoring system. Another example is our prior work on managing routing tables in very large P2P networks =-=[17]-=-. Stavrou et al. suggest management of flash crowd crises by disseminating information over gossip-based unstructured overlays [14]. In [18] we build on top of the research presented in this paper to ...

Citation Context

...o been observed that some of these overlays exhibit properties of small-world and scale-free networks [3, 4]. However, in many cases it is better to construct overlays that are close to random graphs =-=[5]-=-. Gossip-based unstructured overlays can be used as a building block in a variety of network management applications, especially when highly dynamic environments are anticipated. A good example is sel...

Citation Context

...mple [12], there is no need to use a central server. In this light, it is also interesting to mention the recent work on Phenix for the decentralized construction of low-diameter, scale-free networks =-=[19]-=-. In Phenix, an additional goal is to construct networks that are resilient to massive malicious attacks. We have not yet examined this feature for CYCLON, but suspect its good randomness properties w...

Citation Context

...CLON is, in fact, a better candidate than Newscast in that application, as it exhibits a superior randomness property and lower bandwidth requirements 14scompared to Newscast. Finally, in Newscast EM =-=[9]-=-, Kowalczyk and Vlassis use CYCLON for data aggregation, and more specifically as a component of a distributed Expectation-Maximization (EM) algorithm for probabilistic clustering of a set of (geograp...

Citation Context

...cate the document, the client node performs a series of randomized, scoped searches over the links of the overlay created by basic shuffling. In [6], we apply a different gossiping protocol, Newscast =-=[9, 10]-=-, in managing P2P routing tables. In particular, we exploit Newscast’s randomized overlay to pick nodes that satisfy certain criteria, to fill in Pastry-like [11] routing tables. CYCLON is, in fact, a...

Citation Context

...etworks). Unstructured P2P systems are primarily designed to support rapid information dissemination and content-based searching. In this paper we concentrate on gossip-based unstructured P2P systems =-=[6]-=-. These systems aim at exploiting randomness to disseminate information across a large set of nodes. A key issue is to keep the overlay connected even in the event of major disasters without maintaini...