A carregar...
Projeto de investigação
Distributed Data Centric Concurrency Control
Financiador
Autores
Publicações
Data Storage and Dissemination in Pervasive Edge Computing Environments
Publication . Silva, João André Almeida e; Paulino, Hervé; Lourenço, João
Nowadays, smart mobile devices generate huge amounts of data in all sorts of gatherings.
Much of that data has localized and ephemeral interest, but can be of great use if shared
among co-located devices. However, mobile devices often experience poor connectivity,
leading to availability issues if application storage and logic are fully delegated to a
remote cloud infrastructure. In turn, the edge computing paradigm pushes computations
and storage beyond the data center, closer to end-user devices where data is generated
and consumed. Hence, enabling the execution of certain components of edge-enabled
systems directly and cooperatively on edge devices.
This thesis focuses on the design and evaluation of resilient and efficient data storage
and dissemination solutions for pervasive edge computing environments, operating with
or without access to the network infrastructure. In line with this dichotomy, our goal can
be divided into two specific scenarios. The first one is related to the absence of network
infrastructure and the provision of a transient data storage and dissemination system
for networks of co-located mobile devices. The second one relates with the existence of
network infrastructure access and the corresponding edge computing capabilities.
First, the thesis presents time-aware reactive storage (TARS), a reactive data storage
and dissemination model with intrinsic time-awareness, that exploits synergies between
the storage substrate and the publish/subscribe paradigm, and allows queries within a
specific time scope. Next, it describes in more detail: i) Thyme, a data storage and dis-
semination system for wireless edge environments, implementing TARS; ii) Parsley, a
flexible and resilient group-based distributed hash table with preemptive peer relocation
and a dynamic data sharding mechanism; and iii) Thyme GardenBed, a framework
for data storage and dissemination across multi-region edge networks, that makes use of
both device-to-device and edge interactions.
The developed solutions present low overheads, while providing adequate response
times for interactive usage and low energy consumption, proving to be practical in a
variety of situations. They also display good load balancing and fault tolerance properties.
Conflict-Free Replicated Data Types in Dynamic Environments
Publication . Barreto, António José Sá; Paulino, Hervé
Over the years, mobile devices have become increasingly popular and gained improved
computation capabilities allowing them to perform more complex tasks such as
collaborative applications. Given the weak characteristic properties of mobile networks,
which represent highly dynamic environments where users may experience regular involuntary
disconnection periods, the big question arises of how to maintain data consistency.
This issue is most pronounced in collaborative environments where multiple users interact
with each other, sharing a replicated state that may diverge due to concurrency
conflicts and loss of updates.
To maintain consistency, one of today’s best solutions is Conflict-Free Replicated Data
Types (CRDTs), which ensure low latency values and automatic conflict resolution, guaranteeing
eventual consistency of the shared data. However, a limitation often found on
CRDTs and the systems that employ them is the need for the knowledge of the replicas
whom the state changes must be disseminated to. This constitutes a problem since it is
inconceivable to maintain said knowledge in an environment where clients may leave
and join at any given time and consequently get disconnected due to mobile network
communications unreliability.
In this thesis, we present the study and extension of the CRDT concept to dynamic
environments by introducing the developed P/S-CRDTs model, where CRDTs are coupled
with the publisher/subscriber interaction scheme and additional mechanisms to
ensure users are able to cooperate and maintain consistency whilst accounting for the
consequent volatile behaviors of mobile networks. The experimental results show that
in volatile scenarios of disconnection, mobile users in collaborative activity maintain
consistency among themselves and when compared to other available CRDT models, the
P/S-CRDTs model is able to decouple the required knowledge of whom the updates must
be disseminated to, while ensuring appropriate network traffic values.
AtomiS
Publication . Paulino, Hervé; Almeida Matos, Ana; Cederquist, Jan; Giunti, Marco; Matos, João; Ravara, António; NOVALincs; ACM - Association for Computing Machinery
Data-Centric Synchronization (DCS) shifts the reasoning about concurrency restrictions from control structures to data declaration. It is a high-level declarative approach that abstracts away from the actual concurrency control mechanism(s) in use. Despite its advantages, the practical use of DCS is hindered by the fact that it may require many annotations and/or multiple implementations of the same method to cope with differently qualified parameters. To overcome these limitations, in this paper we present AtomiS, a new DCS approach that requires only qualifying types of parameters and return values in interface definitions, and of fields in class definitions. The latter may also be abstracted away in type parameters, rendering class implementations virtually annotation-free. From this high level specification, a static analysis infers the atomicity constraints that are local to each method, considering valid only the method variants that are consistent with the specification, and performs code generation for all valid variants of each method. The generated code is then the target for automatic injection of concurrency control primitives that are responsible for ensuring the absence of data-races, atomicity-violations and deadlocks. We provide a Java implementation and showcase the applicability of AtomiS in real-life code. For the benchmarks analysed, AtomiS requires fewer annotations than the original number of regions requiring locks, as well as fewer annotations than Atomic Sets (a reference DCS proposal).
Key-Value Storage for handling data in mobile devices
Publication . Afonso, José Pedro Serra; Paulino, Hervé
In the current era of technology, computers have shrinked to the point that more than
half of the world population always carries one with them - their mobile devices. These
are used in all sorts of different activities, constantly generating information that needs to
be stored or processed somewhere. To cope with the huge amounts of data generated by
all of these devices, applications have resorted to Cloud services to provide them with the
much needed computational and storage resources, but as these remote infrastructures
still represented a bottleneck communication wise, a new paradigm has been emerging,
Edge Computing. Instead of processing and storing all the data in more distant cloud
services, the data is spread among mobile devices and edge servers connected in a shared
network.
In order to fully take advantage of the low latency times experienced in the Edge,
applications still needed a distributed storage edge-oriented system, capable of handling
the contents generated by all of these mobile devices. The current state-of-the-art storage
systems are able to provide these applications with a storing platform that uses mobile
devices or edge servers as data storing points, but neither uses both.
In this thesis we propose a Key-Value Edge Storage System named Basil, that uses
both mobile devices and edge infrastructures as nodes of the system, capable of providing
users from different locations with a cohesive and consistent distributed storage system.
Furthermore, we will test our KV store against existing NoSQL storage models deployed
in the edge, as well as its own performance while varying the number of nodes it relies
on.
It's about THYME
Publication . Silva, João A.; Cerqueira, Filipe; Paulino, Hervé; Lourenço, João M.; Leitão, João; Preguiça, Nuno; NOVALincs; DI - Departamento de Informática; North-Holland | Elsevier
Nowadays, smart mobile devices generate huge amounts of data in all sorts of gatherings. Much of that data has localized and ephemeral interest, but can be of great use if shared among co-located devices. However, mobile devices often experience poor connectivity, leading to availability issues if application storage and logic are fully delegated to a remote cloud infrastructure. In turn, the edge computing paradigm pushes computations and storage beyond the data center, closer to end-user devices where data is generated and consumed, enabling the execution of certain components of edge-enabled systems directly and cooperatively on edge devices. In this article, we address the challenge of supporting reliable and efficient data storage and dissemination among co-located wireless mobile devices without resorting to centralized services or network infrastructures. We propose THYME, a novel time-aware reactive data storage system for pervasive edge computing environments, that exploits synergies between the storage substrate and the publish/subscribe paradigm. We present the design of THYME and elaborate a three-fold evaluation, through an analytical study, and both simulation and real world experimentations, characterizing the scenarios best suited for its use. The evaluation shows that THYME allows the notification and retrieval of relevant data with low overhead and latency, and also with low energy consumption, proving to be a practical solution in a variety of situations.
Unidades organizacionais
Descrição
Palavras-chave
Contribuidores
Financiadores
Entidade financiadora
Fundação para a Ciência e a Tecnologia
Programa de financiamento
3599-PPCDT
Número da atribuição
PTDC/CCI-COM/32166/2017