Journal article
Alessandro Carrega, Matteo Repetto, Panos Gouvas, Anastasios Zafeiropoulos
IEEE, Cloud Computing (CC), vol. 4(4), 2017 Jul, pp. 26-37
IEEE, Cloud Computing (CC), vol. 4(4), 2017 Jul, pp. 26-37
Cite
Cite
APA
Click to copy
Carrega, A., Repetto, M., Gouvas, P., & Zafeiropoulos, A. (2017). A Middleware for Mobile Edge Computing. Cloud Computing (CC), 4(4), 26β37. https://doi.org/10.1109/MCC.2017.3791021
Chicago/Turabian
Click to copy
Carrega, Alessandro, Matteo Repetto, Panos Gouvas, and Anastasios Zafeiropoulos. βA Middleware for Mobile Edge Computing.β Edited by IEEE. Cloud Computing (CC) 4, no. 4 (July 2017): 26β37.
MLA
Click to copy
Carrega, Alessandro, et al. βA Middleware for Mobile Edge Computing.β Cloud Computing (CC), edited by IEEE, vol. 4, no. 4, July 2017, pp. 26β37, doi:10.1109/MCC.2017.3791021.
BibTeX Click to copy
@article{carrega2017a,
title = {A Middleware for Mobile Edge Computing},
year = {2017},
month = jul,
issue = {4},
journal = {Cloud Computing (CC)},
pages = {26-37},
volume = {4},
doi = {10.1109/MCC.2017.3791021},
author = {Carrega, Alessandro and Repetto, Matteo and Gouvas, Panos and Zafeiropoulos, Anastasios},
editor = {IEEE},
month_numeric = {7}
}
π¬ A Middleware for Mobile Edge Computing: Automated Cross-Domain Orchestration of Distributed Applications
The evolution from centralized cloud computing toward distributed edge infrastructures introduces new challenges in application deployment, resource allocation, mobility support, and lifecycle management.
Our research presents ARCADIA, a middleware framework for developing, deploying, and orchestrating modular applications across heterogeneous Infrastructure-as-a-Service (IaaS) environments, including telecommunications edge installations and traditional cloud data centers.
The proposed solution builds on the ETSI Mobile Edge Computing (MEC) architectural model and integrates three principal components:
πΉ Development Toolkit: Provides a web-based environment for developing microservices, composing service graphs, defining deployment constraints, and specifying lifecycle management policies.
πΉ Smart Controller: Coordinates application placement, resource provisioning, optimization, monitoring, and automated lifecycle management across multiple infrastructure domains.
πΉ Execution Environment: Provides common runtime capabilities, including microservice execution, monitoring, and secure virtual networking between distributed environments.
The framework represents applications as service graphs, where nodes correspond to individual microservices and edges describe their dependencies.
Application metadata define constraints and policies related to resource requirements, locality, QoS, security, scalability, and resilience.
βοΈ Implementation technologies
The experimental implementation combines Java/Spring Boot, OptaPlanner for constraint-based placement optimization, Drools for policy management, ActiveMQ for monitoring communication, and Open Overlay Router for cross-IaaS networking using LISP.
π **Preliminary experimental evaluation
The framework was evaluated using an encrypted Voice-over-IP application composed of five microservices.
β Microservice deployment: Approximately 30 seconds, compared with 60 seconds in the reported comparison.
β Service-graph deployment: Less than 2 minutes for simple graphs, compared with approximately 10 minutes.
β Placement optimization: Less than 30 seconds for the evaluated placement problem.
β Application scaling: Approximately 10 seconds to scale a service graph.
β Development productivity: The graphical composition environment reduced application composition time from days to minutes in the reported preliminary assessment.
β Maintenance: A potential 30% reduction in maintenance time was estimated, rather than experimentally established.
These results are indicative and depend on application complexity and the deployment scenario.
π― Key research contribution: A middleware architecture that integrates application modeling, automated deployment, policy-driven orchestration, and cross-domain resource management to support distributed applications across cloud and mobile edge environments.
The work also identifies open research challenges, particularly seamless cross-domain migration, stronger mobility support, and tighter integration with programmable telecommunications networks.
ππ Bringing the Cloud Closer to Users: The Power of Mobile Edge Computing! βοΈβ‘
What if applications could follow users, run closer to connected devices, and automatically adapt to changing network conditions?
Traditional cloud computing offers enormous computational resources, but centralized data centers can introduce latency, bandwidth limitations, and challenges for highly interactive applications.
Our research introduces ARCADIA, a middleware framework designed to simplify the development, deployment, and management of distributed applications across heterogeneous cloud and edge infrastructures.
The central idea is to make complex infrastructures easier to use through intelligent orchestration.
𧩠Build with microservices: Compose modular applications using reusable software components.
βοΈ Connect cloud and edge: Distribute application components across telecommunications networks, edge installations, and traditional data centers.
βοΈ Automate deployment: Translate application requirements into resource allocation and configuration decisions.
π Manage the lifecycle: Monitor applications and support scaling, reconfiguration, and relocation according to defined policies.
π‘ Move computing closer to users: Support applications requiring geographical proximity, mobility awareness, and low latency.
The manga-inspired infographic illustrates how middleware can transform heterogeneous computing environments into a more integrated and programmable ecosystem.
π― The vision: Applications should focus on delivering services, while intelligent middleware handles the complexity of the underlying infrastructure.
