Middleware for Next Generation Internet Computing9th MW4NG Workshop of the
15th International Middleware Conference 2014
8th MW4NG 2013
7th MW4NG 2012
6th MW4SOC 2011
5th MW4SOC 2010
4th MW4SOC 2009
3rd MW4SOC 2008
2nd MW4SOC 2007
1st MW4SOC 2006
December 8 – 12, 2014
While dependability and security become cornerstones of the information society, they are impaired by change, imprecision, and emerging behavior due to scale, dynamism, and heterogeneity. To address these challenges for next generation Internet computing, key extra-functional properties should not be an "add on" or an "end to end task" anymore, but rather built in by means of Middleware.
Service oriented computing, cloud computing, socio-technical systems, and Web 2.0-style applications are important steps for next generation Internet computing, but still fall short when non functional (a.k.a. extra-functional) quality properties (e.g., dependability, security, performance, and scalability) need to be addressed. The emerging Internet communication architecture (e.g., from projects on the Internet of Things, the Future Internet, etc.) also requires middleware support for delivering computing applications and services. We can see many Internet Computing systems following proprietary end-to-end solutions and being weaved with application-specific approaches. This clearly hinders re-use, which can only be successfully leveraged by Middleware-based solutions. This in turn requires new flexibility for Middleware (adaptivity, elasticity, resilience) and new ways of collaboration between Middleware and applications/services.
Therefore, extra-functional quality properties need to be addressed not only by interfacing and communication standards, but also in terms of actual mechanisms, protocols, and algorithms. Some of the challenges are the administrative heterogeneity, the loose coupling between coarse-grained operations and long-running interactions, high dynamicity, and the required flexibility during run-time. Recently, massive-scale (e.g., big data, millions of participating parties in different roles) and mobility were added to the crucial challenges for Internet computing middleware. The proposed workshop consequently welcomes contributions on how specifically middleware can address the above challenges of next generation Internet computing.
This year, the workshop comprises three tracks: One is the general track on Middleware for Next Generation Internet Computing (MW4NG), then there is the special track on Middleware for Dependability (MidDep), as well as the special track on Middleware for Cyber-Physical Systems (M4CPS). The special tracks have their own track chairs and programme committees, which are detailed below.
09:15–10:30: Session 1 - Middleware for Cyber-Physical Systems (M4CPS)
Miguel Matos (INESC TEC and U. Minho)
Research paper: Towards a Model@Runtime Middleware for Cyber Physical Systems
Francisco Javier Acosta Padilla (INRIA Rennes), Frédéric Weis (IRISA - Université de Rennes 1), and Johann Bourcier (IRISA - Université de Rennes 1)
Cyber Physical Systems (CPS) or Internet of Things systems are typically formed by a myriad of many small interconnected devices. This underlying hardware infrastructure raises new challenges in the way we administrate the software layer of these systems. Indeed, the limited computing power and battery life of each node combined with the very distributed nature of these systems, greatly adds complexity to distributed software layer management. In this paper we propose a new middleware dedicated to CPS to enable the management of software deployment and the dynamic reconfiguration of these systems. Our middleware is inspired from the Component Based Systems and the model@runtime paradigm which has been adapted to the context of Cyber Physical Systems. We have conducted an initial evaluation on a typical Cyber Physical Systems hardware infrastructure which demonstrates the feasibility of providing a model@runtime middleware for these systems.
Research paper: An Architecture for a Smart Spaces Virtual Machine
Leandro Freitas (Universidade Federal de Goiás), Fábio Costa (Universidade Federal de Goiás), Ricardo Rocha (Universidade Federal de Goiás), and Andrew Allen (Georgia Southern University)
The growth and popularization of wireless connectivity and mobile devices have allowed the development of smart spaces that were previously only envisaged in the approach proposed by Mark Weiser. These environments are composed of many computational resources, such as devices and applications, along with user, who must be able to associate with and use these features. However, programming these environments is a challenging task, since smart spaces have a dynamic nature and heterogeneous resources, in addition to the requirement that interactions between users and resources are performed in a coordinated way. We present a new approach for smart spaces programming using Models@RunTime. To this end, we propose a high-level mod- eling language, called 2SML, through which the user can model the smart space with all elements that can be part of it. Models created by the users are interpreted and effected in the physical space by a model execution engine, called 2SVM, whose development is part of this work.
10:30–11:00: Morning Tea/Coffee Break
11:00–12:30: Session 2 - Middleware for Dependability (MidDep)
Research paper: Multi-site Gargamel: Optimistic Synchronization for Reliable Geo-Replicated Databases
Pierpaolo Cincilla, Sébastien Monnet, and Marc Shapiro (Inria and Université Pierre et Marie Curie UPMC - LIP6)
Databases scale poorly in distributed configurations. This is mainly due to the cost of concurrency control and to resource contention. The alternative of centralizing writes works well only for read-intensive workloads, whereas weakening transactional properties is problematic for application developers. In a previous work we introduced Gargamel, a middleware that spreads non-conflicting update transactions to different replicas, but still provides strong transactional guarantees. We extended Gargamel to geo-replication settings. If a datacenter fails, the database remains available at other locations. We minimize the synchronization cost, synchronizing optimistically replicas at distant data-centers. The evaluations of our prototype using distant Amazon datacenters show that Gargamel improves both response time and load by an order of magnitude when contention is high (highly loaded system with bounded resources), and that in the geo-replicated case, the slow-down is negligible.
Research paper: Speculative Client Execution in Deferred Update Replication
Balaji Arun, Sachin Hirve, Roberto Palmieri, Sebastiano Peluso, and Binoy Ravindran (Virginia Tech)
Deferred Update Replication (DUR) is a powerful replication technique that allows parallelism of clients' execution while a global certification phase checks the validity of the transactional execution against workloads running on remote nodes. The well-known favorable scenario of DUR is when remote transactions rarely conflict with each other. In this paper we show that, even in this case, the conflicts happening among local application threads can significantly decrease performance. We address this problem by using speculation. We let local transactions propagate their post-execution snapshot to other local transactions before the outcome of the global certification is notified. This way, in scenarios where accesses are partitioned across nodes, we prevent local transactions from aborting each other. Through experimental study based on well-known transactional benchmarks we assess the effectiveness of the approach, gaining more than 10x using TPC-C benchmark.
Highlight presentation: Geo-Replication with Arbitrary Faults
Marko Vukolic (Eurecom)
Current geo-replicated production systems use crash fault-tolerant (CFT) replication protocols (e.g., Paxos), to maintain consistency despite server crashes and network partitions. These systems do not provide any consistency guarantees as soon as a single non-crash occur, e.g., a bug, data loss or data/memory corruption. Despite years of research on replication protocols handling arbitrary faults (called BFT protocols), these protocols are not used in practice: (i) they require a larger number of servers to tolerate the same number of faults, and (ii) they achieve much lower performance.
In this talk, we present GRAF, a new geo-replication protocol that fills the gap between CFT and BFT. Beyond the guarantees of CFT, GRAF maintains consistency when arbitrary faults occur, with no additional replicas. However, unlike BFT, GRAF tolerates arbitrary faults only when they do not coexist with network partitions. We motivate this design choice by: (i) the fact that arbitrary faults that occur in production are not correlated with network partitions, and (ii) a 3-month study of the connectivity across Amazon EC2 datacenters that shows that network partitions can be considered rare.
We evaluate GRAF using the Amazon EC2 cloud testbed with both microbenchmarks and a real application: the ZooKeeper coordination service. Our experiments show that GRAF performance is almost as good as the best CFT protocol and that it consistently and significantly outperforms BFT protocols. Our GRAF-backed variant of ZooKeeper even outperforms the native crash-tolerant ZooKeeper.
14:15–15:30: Session 3 - Keynote
Keynote: SALTY : A Middleware Toolkit for Building Feedbacks Control Loops
Romain Rouvoy, INRIA (France)
This talk shortly reports on the results of the SALTY R&D project funded by the French funding agency (ANR). The key outcome of this project is a middleware toolkit that covers both design-time and runtime support for integrating self-adaptive behaviours into potentially complex legacy systems. SALTY therefore provides a reflective domain-specific model to externalise and make explicit the control layer of legacy systems. While the adoption of a domain-specific model leverages the mapping on different middleware stacks (FraSCAti or Akka in our case studies), it also acts as a pivot, within a modular toolchain, to implement design-time verifications and to inject runtime guards. Future case studies of this approach will cover green computing, crowd-sensing, and big data systems.
15:30–16:00: Afternoon Tea/Coffee Break
16:00–17:45: Session 4 - Middleware for Next Generation Internet Computing (MW4NG)
Research paper: Towards Efficient and Accurate Privacy Preserving Web Search
Albin Petit (LIRIS), Sonia Ben Mokhtar (LIRIS), Lionel Brunie (LIRIS), and Harald Kosch (University of Passau)
Querying Web search engines is by far the most frequent activity performed by online users and consequently the one in which they are likely to reveal a significant amount of personal information. Protecting the privacy of Web requesters is thus becoming increasingly important. This is often done by using systems that guarantee unlinkability between the requester and her query. The most effective solution to reach this objective is the use of anonymous communication protocols (e.g., onion routing). However, anonymity might not resist to machine learning attacks. Thus, an adversary could link a query to her requester's public profile. Other approaches guarantee unidentifiability of the user interests by generating noise (e.g., creating covert queries or adding extra keywords). However, these solutions overload the network and decrease the accuracy of the results. We present in this paper the first contribution that combines both approaches. It allows a user to perform a private Web search resistant to machine learning attacks while slightly decreasing the relevance of the results. Our three stage architecture contains: (1) a Privacy Proxy that relies on two non-colluding servers to hide the requester identity from the search engine ; (2) a Linkability Assessment that analyses the risk that a request is reassociated with the identity of the requester; (3) an Obfuscator that protects the queries which have been flagged linkable by the linkability assessment.
Research paper: Moana: A Case for Redefining the Internet Service Abstraction
Yan Shvartzshnaider (University of Cambridge) and Max Ott (NICTA)
We introduce Moana, an information-centric middleware service for distributed applications. Moana offers a shared persistent graph-based abstraction through which applications can communicate with each other by extending and observing the shared graph. This paper describes the Moana service model as well as discusses the design goals and challenges for such a middleware to be a viable and competitive alternative to current practice.
Research paper: Distributed Event-Based System with Multiscoping for Multiscalability
Léon Lim (Institut Mines Télécom) and Denis Conan (Institut Mines Télécom)
Distributed Event-Based System (DEBS) provides a versatile solution for asynchronously exchanging data in a distributed system, loosely-coupled in space and time. The software architecture of a DEBS is composed of an overlay network of brokers that are responsible for routing data from producers to consumers. An important issue is the cost (in terms of exchanged messages) of the installation of advertisement or subscription filters on the brokers and the cost of routing notifications. The problem is exacerbated in large and heterogeneous systems involving clouds, cloudlets, desktops, laptops, mobile phones, and smart objects of the Internet of Things (IoT). In this paper, we associate the system concept of scale (of multiscale distributed systems) with the concept of scope (of DEBS) and we introduce DEBS with multiscoping. We also extend the requirements of distributed routing to deal with multiscoping. In the context of the IoT, we show in an illustrative example that the solution allows application designers and system administrators to tag advertisements and subscriptions for semantically delimiting scopes that are superposed.
Miguel Matos (INESC TEC and U. Minho)
Miguel Matos (main contact)
High Assurance Software Lab, INESC TEC and U. Minho, Portugal
Phone: +351 253 604 477
Karl M. Göschka
Vienna University of Technology, Austria
Patrick C. K. Hung
Faculty of Business and IT, University of Ontario Institute of Technology, Canada
This workshop has its own ISBN and will be included in the ACM digital library.