Anne-Marie Kermarrec

The many faces of publish/subscribe

Well adapted to the loosely coupled nature of distributed interaction in large-scale applications, the publish/subscribe communication paradigm has recently received increasing attention. With systems based on the publish/subscribe interaction scheme, subscribers register their interest in an event, or a pattern of events, and are subsequently asynchronously notified of events generated by publishers. Many variants of the paradigm have recently been proposed, each variant being specifically adapted to some given application or network model. This paper factors out the common denominator underlying these variants: full decoupling of the communicating entities in time, space, and synchronization. We use these three decoupling dimensions to better identify commonalities and divergences with traditional interaction paradigms. The many variations on the theme of publish/subscribe are classified and synthesized. In particular, their respective benefits and shortcomings are discussed both in terms of interfaces and implementations.

Scribe: a large-scale and decentralized application-level multicast infrastructure

This paper presents Scribe, a scalable application-level multicast infrastructure. Scribe supports large numbers of groups, with a potentially large number of members per group. Scribe is built on top of Pastry, a generic peer-to-peer object location and routing substrate overlayed on the Internet, and leverages Pastrys reliability, self-organization, and locality properties. Pastry is used to create and manage groups and to build efficient multicast trees for the dissemination of messages to each group. Scribe provides best-effort reliability guarantees, and we outline how an application can extend Scribe to provide stronger reliability. Simulation results, based on a realistic network topology model, show that Scribe scales across a wide range of groups and group sizes. Also, it balances the load on the nodes while achieving acceptable delay and link stress when compared with Internet protocol multicast.

SplitStream: high-bandwidth multicast in cooperative environments

In tree-based multicast systems, a relatively small number of interior nodes carry the load of forwarding multicast messages. This works well when the interior nodes are highly-available, dedicated infrastructure routers but it poses a problem for application-level multicast in peer-to-peer systems. SplitStream addresses this problem by striping the content across a forest of interior-node-disjoint multicast trees that distributes the forwarding load among all participating peers. For example, it is possible to construct efficient SplitStream forests in which each peer contributes only as much forwarding bandwidth as it receives. Furthermore, with appropriate content encodings, SplitStream is highly robust to failures because a node failure causes the loss of a single stripe on average. We present the design and implementation of SplitStream and show experimental results obtained on an Internet testbed and via large-scale network simulation. The results show that SplitStream distributes the forwarding load among all peers and can accommodate peers with different bandwidth capacities while imposing low overhead for forest construction and maintenance.

SCRIBE: The Design of a Large-Scale Event Notification Infrastructure

This paper presents Scribe, a large-scale event notification infrastructure for topic-based publish-subscribe applications. Scribe supports large numbers of topics, with a potentially large number of subscribers per topic. Scribe is built on top of Pastry, a generic peer-to-peer object location and routing substrate overlayed on the Internet, and leverages Pastrys reliability, self-organization and locality properties. Pastryi s used to create a topic (group) and to build an efficient multicast tree for the dissemination of events to the topics subscribers (members). Scribe provides weak reliability guarantees, but we outline how an application can extend Scribe to provide stronger ones.

Peer-to-peer membership management for gossip-based protocols

Gossip-based protocols for group communication have attractive scalability and reliability properties. The probabilistic gossip schemes studied so far typically assume that each group member has full knowledge of the global membership and chooses gossip targets uniformly at random. The requirement of global knowledge impairs their applicability to very large-scale groups. In this paper, we present SCAMP (Scalable Membership protocol), a novel peer-to-peer membership protocol which operates in a fully decentralized manner and provides each member with a partial view of the group membership. Our protocol is self-organizing in the sense that the size of partial views naturally converges to the value required to support a gossip algorithm reliably. This value is a function of the group size, but is achieved without any node knowing the group size. We propose additional mechanisms to achieve balanced view sizes even with highly unbalanced subscription patterns. We present the design, theoretical analysis, and a detailed evaluation of the basic protocol and its refinements. Simulation results show that the reliability guarantees provided by SCAMP are comparable to previous schemes based on global knowledge. The scale of the experiments attests to the scalability of the protocol.

Epidemic information dissemination in distributed systems

Easy to deploy, robust, and highly resilient to failures, epidemic algorithms are a potentially effective mechanism for propagating information in large peer-to-peer systems deployed on Internet or ad hoc networks. It is possible to adjust the parameters of epidemic algorithm to achieve high reliability despite process crashes and disconnections, packet losses, and a dynamic network topology. Although researchers have used epidemic algorithms in applications such as failure detection, data aggregation, resource discovery and monitoring, and database replication, their general applicability to practical, Internet-wide systems remains open to question. We describe four key problems: membership maintenance, network awareness, buffer management, and message filtering, and suggest some preliminary approaches to address them.

Lightweight probabilistic broadcast

Gossip-based broadcast algorithms, a family of probabilistic broadcast algorithms, trade reliability guarantees against scalability properties. Scalability in this context has usually been expressed in terms of message throughput and delivery latency, but there has been little work on how to reduce the memory consumption for membership management and message buffering at large scale.This paper presents lightweight probabilistic broadcast (lpbcast), a novel gossip-based broadcast algorithm, which complements the inherent throughput scalability of traditional probabilistic broadcast algorithms with a scalable memory management technique. Our algorithm is completely decentralized and based only on local information: in particular, every process only knows a fixed subset of processes in the system and only buffers fixed most suitable subsets of messages. We analyze our broadcast algorithm stochastically and compare the analytical results both with simulations and concrete implementation measurements.

SplitStream: High-Bandwidth Content Distribution in Cooperative Environments

In tree-based multicast systems, a relatively small number of interior nodes carry the load of forwarding multicast messages. This works well when the interior nodes are dedicated infrastructure routers. But it poses a problem in cooperative application-level multicast, where participants expect to contribute resources proportional to the benefit they derive from using the system. Moreover, many participants may not have the network capacity and availability required of an interior node in high-bandwidth multicast applications. SplitStream is a high-bandwidth content distribution system based on application-level multicast. It distributes the forwarding load among all the participants, and is able to accommodate participating nodes with different bandwidth capacities. We sketch the design of SplitStream and present some preliminary performance results.

Probabilistic reliable dissemination in large-scale systems

The growth of the Internet raises new challenges for the design of distributed systems and applications. In the context of group communication protocols, gossip-based schemes have attracted interest as they are scalable, easy to deploy, and resilient to network and process failures. However, traditional gossip-based protocols have two major drawbacks: 1) they rely on each peer having knowledge of the global membership; and 2) being oblivious to the network topology, they can impose a high load on network links when applied to wide-area settings. In this paper, we provide a theoretical analysis of gossip-based protocols which relates their reliability to key system parameters (the system size, failure rates, and number of gossip targets). The results provide guidelines for the design of practical protocols. In particular, they show how reliability can be maintained while alleviating drawback by: 1) providing each peer with only a small subset of the total membership information and drawback; and 2) organizing members into a hierarchical structure that reflects their proximity according to some network-related metric. We validate the analytical results by simulations and verify that the hierarchical gossip protocol considerably reduces the load on the network compared to the original, non-hierarchical protocol.

An evaluation of scalable application-level multicast built using peer-to-peer overlays

Structured peer-to-peer overlay networks such as CAN, Chord, Pastry, and Tapestry can be used to implement Internet-scale application-level multicast. There are two general approaches to accomplishing this: tree building and flooding. This paper evaluates these two approaches using two different types of structured overlay: 1) overlays which use a form of generalized hypercube routing, e.g., Chord, Pastry and Tapestry, and 2) overlays which use a numerical distance metric to route through a Cartesian hyperspace, e.g., CAN. Pastry and CAN are chosen as the representatives of each type of overlay. To the best of our knowledge, this paper reports the first head-to-head comparison of CAN-style versus Pastry-style overlay networks, using multicast communication workloads running on an identical simulation infrastructure. The two approaches to multicast are independent of overlay network choice, and we provide a comparison of flooding versus tree-based multicast on both overlays. Results show that the tree-based approach consistently outperforms the flooding approach. Finally, for tree-based multicast, we show that Pastry provides better performance than CAN.