Publications
129 results found
Novotny P, Ko B-J, Wolf AL, 2018, Locating faults in MANET-hosted software systems, IEEE Transactions on Dependable and Secure Systems, Vol: 15, Pages: 452-465, ISSN: 1545-5971
We present a method to locate faults in service-based software systems hosted on mobile ad hoc networks (MANETs). Insuch systems, computations are structured as interdependent services distributed across the network, collaborating to satisfy clientrequests. Faults, which may occur at either or both the service and network layers, propagate by cascading through some subset of theservices, from their root causes back to the clients that initiate requests. Fault localization in this environment is especially challengingbecause the systems are typically subject to a wider variety and higher incidence of faults than those deployed in fixed networks, theresources available to collect and store analysis data are severely limited, and many of the sources of faults are by their naturetransient. Our method makes use of service-dependence and fault data that are harvested in the network through decentralized,run-time observations of service interactions and fault symptoms. We have designed timing- and Bayesian-based reasoningtechniques to analyze the data in the context of a specific fault propagation model. The analysis provides a ranked list of candidate faultlocations. Through extensive simulations, we evaluate the performance of our method in terms of its accuracy in correctly ranking rootcauses under a wide range of operational conditions.
Silvestri S, Holbert B, Novotny P, et al., 2017, Inferring Network Topologies in MANETs: Application to Service Redeployment, ADVANCES IN COMPUTER COMMUNICATIONS AND NETWORKS: FROM GREEN, MOBILE, PERVASIVE NETWORKING TO BIG DATA COMPUTING, Editors: Sha, Striegel, Song, Publisher: RIVER PUBLISHERS, Pages: 127-154, ISBN: 978-87-93379-87-9
Yu T, Feng B, Stillwell M, et al., 2016, Relation-oriented resource allocation for multi-accelerator systems, 27th IEEE International Conference on Application-specific Systems, Architectures and Processors (ASAP), Publisher: IEEE, Pages: 243-244, ISSN: 2160-052X
This paper presents a novel approach for allocating resources in systems with multiple accelerators. It has three main contributions. First, a new model based on Birkhoff's representation theory in capturing the ordering properties of resource allocation requests (RArs). Second, an effective technique for resource allocation based on this model, targeting systems with multiple accelerators. Third, the evaluation of the proposed approach for Maxeler MPC-X multi-accelerator systems, demonstrating time-efficiency and 30%-50% failure-rate decrease (FRD) on random input dataset.
Pietzuch PR, Koliousis A, Weidlich M, et al., 2016, Demo- The SABER system for window-based hybrid stream processing with GPGPUs, DEBS 2016, Publisher: Association for Computing Machinery, Pages: 354-357
Heterogeneous architectures that combine multi-core CPUs withmany-core GPGPUs have the potential to improve the performanceof data-intensive stream processing applications. Yet, a stream pro-cessing engine must execute streaming SQL queries with sufficientdata-parallelism to fully utilise the available heterogeneous proces-sors, and decide how to use each processor in the most effectiveway. Addressing these challenges, we demonstrate SABER, ahybrid high-performance relational stream processing engine forCPUs and GPGPUs. SABER executes window-based streaming SQL queries in a data-parallel fashion and employs an adaptive scheduling strategy to balance the load on the different types of processors. To hidedata movement costs, SABER pipelines the transfer of stream databetween CPU and GPGPU memory. In this paper, we review thedesign principles of SABER in terms of its hybrid stream processingmodel and its architecture for query execution. We also present aweb front-end that monitors processing throughput.
Alim A, Clegg RG, Mai L, et al., 2016, FLICK: developing and running application-specific network services, 2016 USENIX Annual Technical Conference (USENIX ATC '16), Publisher: USENIX Association
Data centre networks are increasingly programmable,with application-specific network services proliferating,from custom load-balancers to middleboxes providingcaching and aggregation. Developers must currently implementthese services using traditional low-level APIs,which neither support natural operations on applicationdata nor provide efficient performance isolation.We describe FLICK, a framework for the programmingand execution of application-specific network serviceson multi-core CPUs. Developers write network servicesin the FLICK language, which offers high-level processingconstructs and application-relevant data types.FLICK programs are translated automatically to efficient,parallel task graphs, implemented in C++ on top of auser-space TCP stack. Task graphs have bounded resourceusage at runtime, which means that the graphsof multiple services can execute concurrently withoutinterference using cooperative scheduling. We evaluateFLICK with several services (an HTTP load-balancer,a Memcached router and a Hadoop data aggregator),showing that it achieves good performance while reducingdevelopment effort.
Koliousis A, Weidlich M, Fernandez R, et al., 2016, Saber: window-based hybrid stream processing for heterogeneous architectures, 2016 ACM SIGMOD/PODS Conference, Publisher: ACM, Pages: 555-569
Modern servers have become heterogeneous, often combining multicoreCPUs with many-core GPGPUs. Such heterogeneous architectureshave the potential to improve the performance of data-intensivestream processing applications, but they are not supported by currentrelational stream processing engines. For an engine to exploit aheterogeneous architecture, it must execute streaming SQL querieswith sufficient data-parallelism to fully utilise all available heterogeneousprocessors, and decide how to use each in the most effectiveway. It must do this while respecting the semantics of streamingSQL queries, in particular with regard to window handling.We describe SABER, a hybrid high-performance relational streamprocessing engine for CPUs and GPGPUs. SABER executes windowbasedstreaming SQL queries in a data-parallel fashion using allavailable CPU and GPGPU cores. Instead of statically assigningquery operators to heterogeneous processors, SABER employs anew adaptive heterogeneous lookahead scheduling strategy, whichincreases the share of queries executing on the processor that yieldsthe highest performance. To hide data movement costs, SABERpipelines the transfer of stream data between different memory typesand the CPU/GPGPU. Our experimental comparison against state-ofthe-artengines shows that SABER increases processing throughputwhile maintaining low latency for a wide range of streaming SQLqueries with small and large windows sizes.
Wolf AL, 2016, Moving Forward, Communications of the ACM, Vol: 59, Pages: 5-5, ISSN: 0001-0782
Novotny P, Wolf A, 2016, Simulating services-based systems hosted in networks with dynamic topology, Departmental Technical Report: 16/2, Publisher: Department of Computing, Imperial College London, 16/2
The emerging use of mobile ad hoc networks combined with current trends in the use ofservice-based systems pose new challenges to accurate simulation of these systems. Currentnetwork simulators lack the ability to replicate the complex message exchange behaviour ofservices, while service simulators do not accurately capture of mobile network properties. Inthis paper we provide an overview of a framework for simulating both a service behaviouralmodel and a mobile network. The framework is implemented as an extension of the NS-3network simulator.
Wolf AL, 2015, A New Chief Executive Officer and Executive Director of ACM, COMMUNICATIONS OF THE ACM, Vol: 58, Pages: 4-5, ISSN: 0001-0782
Novotny P, Ko BJ, Wolf AL, 2015, On-Demand Discovery of Software Service Dependencies in MANETs, IEEE TRANSACTIONS ON NETWORK AND SERVICE MANAGEMENT, Vol: 12, Pages: 278-292, ISSN: 1932-4537
- Author Web Link
- Cite
- Citations: 6
Pamboris A, Báguena M, Wolf AL, et al., 2015, Demo: NOMAD: an edge cloud platform for hyper-responsive mobile apps, 13th Annual International Conference on Mobile Systems, Applications, and Services (MobiSys 2015), Publisher: Association for Computing Machinery, Pages: 459-459
Foukas X, Carzaniga A, Wolf AL, 2015, Measuring the Mixing Time of a Network, 34th IEEE Conference on Computer Communications (INFOCOM), Publisher: IEEE, ISSN: 0743-166X
Silvestri S, Holbert B, Novotny P, et al., 2015, Inferring Network Topologies in MANETs Applied to Service Redeployment, International Conference on Computer Communications and Networks ICCCN, Publisher: IEEE, ISSN: 1095-2055
Novotny P, Urgaonkar R, Wolf AL, et al., 2015, Dynamic Placement of Composite Software Services in Hybrid Wireless Networks, 34th IEEE Annual Military Communications Conference (MILCOM) on Leveraging Technology - The Joint Imperative, Publisher: IEEE, Pages: 1052-1057, ISSN: 2155-7578
- Author Web Link
- Cite
- Citations: 1
Novotny P, Ko BJ, Wolf AL, 2015, Delay Tolerant Harvesting of Monitoring Data for MANET-Hosted Service-Based Systems, 12th IEEE International Conference on Services Computing (SCC), Publisher: IEEE, Pages: 9-16
- Author Web Link
- Cite
- Citations: 1
Pietzuch PR, Mai L, Rupprecht L, et al., 2014, NetAgg: Middleboxes for Application-Specific Traffic Aggregation, 10th International Conference on Emerging Networking Experiments and Technologies (CoNEXT), Publisher: ACM, Pages: 249-262
Carzaniga A, Wolf AL, 2014, IEEE Conference on Computer Communications (INFOCOM), IEEE Conference on Computer Communications (INFOCOM)
Papalini M, Carzaniga A, Khazaei K, et al., 2014, Scalable routing for tag-based information-centric networking, Pages: 17-26
Routing in information-centric networking remains an open problem. The main issue is scalability. Traditional IP routing can be used with name prefixes, but it is believed that the number of prefixes will grow too large. A related problem is the use of per-packet in-network state (to cut loops and return data to consumers). We develop a routing scheme that solves these problems. The service model of our information-centric network supports information pull and push using tag sets as information descriptors. Within this service model, we propose a routing scheme that supports forwarding along multiple loop-free paths, aggregates addresses for scalability, does not require per-packet network state, and leads to near-optimal paths on average. We evaluate the scalability of our routing scheme, both in terms of memory and computational complexity, on the full Internet AS-level topology and on the internal networks of representative ASes using realistic distributions of content and users extrapolated from traces of popular applications. For example, a population of 500 million users requires a routing information base of 3.8GB with an almost at growth and, in this case, a routing update (one content descriptor) can be processed in 2ms on commodity hardware. We conclude that information-centric networking is feasible, even with (or perhaps thanks to) addresses consisting of expressive content descriptors.
Mai L, Rupprecht L, Costa P, et al., 2013, Supporting Application-specific In-network Processing in Data Centres, ACM SIGCOMM COMPUTER COMMUNICATION REVIEW, Vol: 43, Pages: 519-520, ISSN: 0146-4833
- Author Web Link
- Cite
- Citations: 3
Carzaniga A, Khazaei K, Papalini M, et al., 2013, Is Information-Centric Multi-Tree Routing Feasible?, ACM SIGCOMM COMPUTER COMMUNICATION REVIEW, Vol: 43, ISSN: 0146-4833
- Author Web Link
- Cite
- Citations: 7
Carzaniga A, Khazaei K, Papalini M, et al., 2013, Is information-centric multi-tree routing feasible?, Pages: 3-8
We have argued that an information-centric network should natively support publish/subscribe event notification in addition to on-demand content delivery. We have also argued that both primitives could use the same forwarding information base and, furthermore, that both primitives can easily support addresses that are more expressive than simple hierarchical names. In this paper we present a concrete routing scheme that realizes this: "push" as well as "pull" communication; anycast as well as multicast; and descriptor-based (as opposed to name-based) addressing. The scheme is founded on multiple tree covers that can be arranged and composed hierarchically following the structure of network domains. On each tree, the scheme combines addresses so as to reduce forwarding state. We demonstrate the feasibility and scalability of the scheme through simulations on Internet-scale workloads in realistic network settings. Copyright © 2013 ACM.
Mai L, Rupprecht L, Costa P, et al., 2013, Supporting application-specific in-network processing in data centres, SIGCOMM 2013 - Proceedings of the ACM SIGCOMM 2013 Conference on Applications, Technologies, Architectures, and Protocols for Computer Communication, Pages: 519-520
- Cite
- Citations: 1
Novotny P, Wolf AL, Ko BJ, 2013, Discovering Service Dependencies in Mobile Ad Hoc Networks, 13th IFIP/IEEE International Symposium on Integrated Network Management (IM), Publisher: IEEE, Pages: 527-533
- Author Web Link
- Cite
- Citations: 5
Tati S, Novotny P, Ko BJ, et al., 2013, Diagnosing Degradation of Services in Hybrid Wireless Tactical Networks, Conference on Ground/Air Multisensor Interoperability, Integration, and Networking for Persistent ISR IV, Publisher: SPIE-INT SOC OPTICAL ENGINEERING, ISSN: 0277-786X
Novotny P, Wolf AL, Ko B-J, 2012, Fault Localization in MANET-Hosted Service-Based Systems
Novotny P, Ko BJ, Wolf A, 2012, Techniques for locating service faults in mobile ad hoc networks, Departmental Technical Report: 12/3, Publisher: Department of Computing, Imperial College London, 12/3
Fault localization in general refers to a technique for identifying the likely root causes offailures observed in systems formed from components. Fault localization in systems deployed onmobile ad hoc networks (MANETs) is a particularly challenging task because those systems aresubject to a wider variety and higher incidence of faults than those deployed in xed networks,the resources available to track fault symptoms are severely limited, and many of the sources offaults in MANETs are by their nature transient. We present a method for localizing the faultsoccurring in service-based systems hosted on MANETs. The method is based on the use ofdependence data that are discovered dynamically through decentralized observations of serviceinteractions. We employ both Bayesian and timing-based reasoning techniques to analyze thedata in the context of a speci c fault propagation model, deriving a ranked list of candidatefault locations. We present the results of an extensive set of experiments exploring a widerange of operational conditions to evaluate the accuracy of our method.
Costa P, Migliavacca M, Pietzuch P, et al., 2012, NaaS: Network-as-a-Service in the Cloud, 2nd USENIX Workshop on Hot Topics in Management of Internet, Cloud, and Enterprise Networks and Services
Novotny P, Ko BJ, Wolf A, 2012, Discovering service dependencies in mobile ad hoc networks, Departmental Technical Report: 12/2, Publisher: Department of Computing, Imperial College London, 12/2
The combination of service-oriented applications, with their run-time service binding, andmobile ad hoc networks, with their transient communication topologies, brings a new levelof complex dynamism to the structure and behavior of software systems. This complexitychallenges our ability to understand the dependence relationships among system componentswhen performing analyses such as fault localization and impact analysis. Current methods ofdynamic dependence discovery, developed for use in xed networks, assume that dependencieschange slowly. Moreover, they require relatively long monitoring periods as well as substantialmemory and communication resources, which are impractical in the mobile ad hoc networkenvironment. We describe a new method, designed speci cally for this environment, thatallows the engineer to trade accuracy against cost, yielding dynamic snapshots of dependencerelationships. Through extensive simulations, we evaluate the performance of our method interms of the accuracy of the discovered dependencies, and draw insights on the selection ofcritical parameters under various operational conditions.
Carzaniga A, Papalini M, Wolf AL, 2011, Content-based publish/subscribe networking and information-centric networking, Pages: 56-61
On-line information comes in different forms and is accessed in different ways and for different purposes. For example, a recording of Beethoven's Ninth Symphony differs from a storm warning from the local weather service. Beethoven's Ninth is a large media file with perpetual validity that is typically accessed on demand by users. By contrast, a storm warning is a small ephemeral message typically pushed by the weather service to all users in a specific geographic area. We argue that both should and would be well supported by an information-centric network. More specifically we argue three points. First, modern applications, reflecting the nature of human communications, use and transmit large and long-lived files as well as small ephemeral messages. Second, accessing those two types of information involves significantly different operations within the network. Third, despite their differences, both types of information would benefit from an addressing scheme based on content rather than on more or less flat identifiers, which means that both should be integrated to some extent within a unified content-based routing infrastructure. © 2011 ACM.
This data is extracted from the Web of Science and reproduced under a licence from Thomson Reuters. You may not copy or re-distribute this data in whole or in part without the written consent of the Science business of Thomson Reuters.