Roman Vitenberg

Group communication specifications: a comprehensive study

View-oriented group communication is an important and widely usedbuilding block for many distributed applications. Much currentresearch has been dedicated to specifying the semantics andservices of view-oriented group communication systems (GCSs).However, the guarantees of different GCSs are formulated usingvarying terminologies and modeling techniques, and thespecifications vary in their rigor. This makes it difficult toanalyze and compare the different systems. This survey provides acomprehensive set of clear and rigorous specifications, which maybe combined to represent the guarantees of most existing GCSs. Inthe light of these specifications, over 30 published GCSspecifications are surveyed. Thus, the specifications serve as aunifying framework for the classification, analysis, and comparisonof group communication systems. The survey also discusses over adozen different applications of group communication systems,shedding light on the usefulness of the presented specifications.This survey is aimed at both system builders and theoreticalresearchers. The specification framework presented in this articlewill help builders of group communication systems understand andspecify their service semantics; the extensive survey will allowthem to compare their service to others. Application builders willfind a guide here to the services provided by a large variety ofGCSs, which could help them choose the GCS appropriate for theirneeds. The formal framework may provide a basis for interestingtheoretical work, for example, analyzing relative strengths ofdifferent properties and the costs of implementing them.

SpiderCast: a scalable interest-aware overlay for topic-based pub/sub communication

We introduce SpiderCast, a distributed protocol for constructing scalable churn-resistant overlay topologies for supporting decentralized topic-based pub/sub communication. SpiderCast is designed to effectively tread the balance between average overlay degree and communication cost of event dissemination. It employs a novel coverage-optimizing heuristic in which the nodes utilize partial subscription views (provided by a decentralized membership service) to reduce the average node degree while guaranteeing (with high probability) that the events posted on each topic can be routed solely through the nodes interested in this topic (in other words, the overlay is topic-connected). SpiderCast is unique in maintaining an overlay topology that scales well with the average number of topics a node is subscribed to, assuming the subscriptions are correlated insofar as found in most typical workloads. Furthermore, the degree grows logarithmically in the total number of topics, and slowly decreases as the number of nodes increases. We show experimentally that, for many practical work-loads, the SpiderCast overlays are both topic-connected and have a low per-topic diameter while requiring each node to maintain a low average number of connections. These properties are satisfied even in very large settings involving up to 10,000 nodes, 1,000 topics, and 70 subscriptions per-node, and under high churn rates. In addition, our results demonstrate that, in a large setting, the average node degree in SpiderCast is at least 45% smaller than in other overlays typically used to support decentralized pub/sub communication (such as e.g., similarity-based, rings-based, and random overlays).

PolderCast: fast, robust, and scalable architecture for P2P topic-based pub/sub

We propose PolderCast, a P2P topic-based Pub/Sub system that is (a) fault-tolerant and robust, (b) scalable w.r.t the number of nodes interested in a topic and number of topics that nodes are interested in, and (c) fast in terms of dissemination latency while (d) attaining a low communication overhead. This combination of properties is provided by an implementation that blends deterministic propagation over maintained rings with probabilistic dissemination following a limited number of random shortcuts. The rings are constructed and maintained using gossiping techniques. The random shortcuts are provided by two distinct peer-sampling services: Cyclon generates purely random links while Vicinity produces interest-induced random links. We analyze PolderCast and survey it in the context of existing approaches. We evaluate PolderCast experimentally using real-world workloads from Twitter and Facebook traces. We use widely renowned Scribe [5] as a baseline in a number of experiments. Robustness with respect to node churn is evaluated through traces from the Skype superpeer network. We show that the experimental results corroborate all of the above properties in settings of up to 10K nodes, 10K topics, and 5K topics per-node.

Adaptable service composition for very-large-scale Internet of Things systems

The (future) Internet of Things (IoT), with service composition point of view, raises additional challenges especially with respect to handling the scale, dynamicity and heterogeneity of the target networking environment. Therefore, the services offered by IoT resources can not be composed by simply extending existing Service Oriented Architecture (SOA) approaches, since it requires the integration of a huge number of real world services that demands for the user-centric and a situation-aware composition process. In this paper, we present an architectural approach that enables efficient and adaptive composition of services by locally orchestrating the distributed web-enabled services in Very Large Scale (VLS) IoT systems and globally choreographing the different enterprise level Web services.

Divide and Conquer Algorithms for Publish/Subscribe Overlay Design

Overlay network design for topic-based publish/subscribe systems is of primary importance because the overlay directly impacts the systems performance. Determining a topic-connected overlay, in which for every topic the graph induced by nodes interested in the topic is connected, is a fundamental problem. Existing algorithms for this problem suffer from three key drawbacks: (1) prohibitively high running time cost, (2) requirement of full system knowledge and centralized operation, and (3) constructing overlay from scratch. From a practical point of view, these are all significant limitations. To address these concerns, in this paper, we develop novel algorithms that efficiently solve the problem of dynamically joining two or more topic-connected overlays. Inspired from the divide-and-conquer character of our approach, we derive an algorithm that solves the original problem at a fraction (up to 1.7\%) of the running time cost of alternative solutions, but at the expense of an empirically insignificant increase in the average node degree.

On the composability of consistency conditions

This paper presents a formal framework, which is based on the notion of a serialization set , that enables to compose a set of consistency conditions into a more restrictive one. To exemplify the utility of this framework, a list of very basic consistency conditions is identified, and it is shown that various compositions of the basic conditions yield some of the most commonly used consistency conditions, such as sequential consistency, causal memory , and Pipelined RAM. The paper also lists several applications that can benefit from even weaker semantics than Pipelined RAM that can be expressed as a composition of a small subset of the basic conditions.

Consistency Conditions for a CORBA Caching Service

Distributed object caching is essential for building and deploying Internet wide services based on middlewares such as CORBA. By caching objects, it is possible to mask much of the latency associated with accessing remote objects, to provide more predictable quality of service to clients, and to improve the scalability of the service. This paper presents a combined theoretical and practical view on specifying and implementing consistency conditions for such a service. First, a formal definition of a set of basic consistency conditions is given in an abstract, implementation independent manner. It is then shown that common consistency conditions such as sequential consistency, causal consistency, and PRAM can be formally specified as a combination of these more basic conditions. Finally, the paper describes the implementation of the proposed basic consistency conditions in CASCADE, a distributed CORBA object caching service.

Cost-Effective Resource Allocation for Deploying Pub/Sub on Cloud

Publish/subscribe (pub/sub) is a popular communication paradigm in the design of large-scale distributed systems. A fundamental challenge in deploying pub/sub systems on a data center or a cloud infrastructure is efficient and cost-effective resource allocation that would allow delivery of notifications to all subscribers. In this paper, we provide answers to the following three fundamental questions: Given a pub/sub workload, (1) what is the minimum amount of resources needed to satisfy all the subscribers, (2) what is a cost-effective way to allocate resources for the given workload, and (3) what is the cost of hosting it on a public Infrastructure-as-a-Service (IaaS) provider like Amazon EC2. To answer these questions, we formulate a problem coined Minimum Cost Subscriber Satisfaction (MCSS). We prove MCSS to be NP-hard and provide an efficient heuristic solution based on a combination of optimizations. We evaluate the solution experimentally using real traces from Spotify and Twitter along with a pricing model from Amazon. We show the impact of each optimization using a naive solution as the baseline. Using a variety of practical scenarios for each dataset, we also show that our solution scales well for millions of subscribers and runs fast.

Distributed event aggregation for content-based publish/subscribe systems

Modern data-intensive applications handling massive event streams such as real-time traffic monitoring require support for both rich data filtering and aggregation. While the pub/sub communication paradigm provides an effective solution for the sought semantic diversity of event filtering, the event processing capabilities of existing pub/sub systems are restricted to singular event matching without support for stream aggregation, which so far can be accommodated only at the subscriber edge brokers. In this paper, we propose the first systematic solution for supporting distributed aggregation over a range of time-based aggregation window semantics in a content-based pub/sub system. In order to eschew the need to disseminate a large number of publications to subscribers, we strive to distribute the aggregation computation within the pub/sub overlay network. By enriching the pub/sub language with aggregation semantics, we allow pub/sub brokers to aggregate incoming publications and forward only results to the next broker downstream. We show that our baseline solutions, one which aggregates early (at the publisher edge) and another which aggregates late (at the subscriber edge), are not optimal strategies for minimizing bandwidth consumption. We then propose an adaptive rate-based heuristic solution which determines which brokers should aggregate publications. Using real datasets extracted from our traffic monitoring use case, we show that this adaptive solution leads to improved performance compared to that of our baseline solutions.

A generalized algorithm for publish/subscribe overlay design and its fast implementation

It is a challenging and fundamental problem to construct the underlying overlay network to support efficient and scalable information distribution in topic-based publish/subscribe systems. Existing overlay design algorithms aim to minimize the node fan-out while building topic-connected overlays, in which all nodes interested in the same topic are organized in a directly connected dissemination sub-overlay. However, most state-of-the-art algorithms suffer from high computational complexity, such as O(|V|4|T|), where V is the node set and T is the topic set. We devise a general indexing data structure that provides a significantly faster implementation, with O(|V|2|T|) running time, for different state-of-the-art algorithms. The generality of the indexing data structure is due to the fact that it enables edge lookup by both node degree and edge contribution, a central metric in all existing algorithms. When tested on typical pub/sub workloads, the speedup observed was by a factor of over 1 000, thereby rendering the algorithms more suitable for practical use. For example, under a typically Zipf distributed pub/sub workload, with 1 000 nodes and 100 topics, our new implementation completes in 3.823 seconds, while the previous alternative takes over 555 minutes.