An ensemble online learning-based approach for VNF scaling in the Edge Computing
Edge Computing (EC) platforms have emerged as essential solutions for managing applications with high computational demands and strict response time requirements. These platforms capitalize on the decentralized nature of edge devices, which are situated close to end-users and data sources, thus mi...
Na minha lista:
| Autor principal: | |
|---|---|
| Outros Autores: | |
| Formato: | doctoralThesis |
| Idioma: | pt_BR |
| Publicado em: |
Universidade Federal do Rio Grande do Norte
|
| Assuntos: | |
| Endereço do item: | https://repositorio.ufrn.br/handle/123456789/56544 |
| Tags: |
Adicionar Tag
Sem tags, seja o primeiro a adicionar uma tag!
|
| Resumo: | Edge Computing (EC) platforms have emerged as essential solutions for managing applications with high computational demands and strict response time requirements. These
platforms capitalize on the decentralized nature of edge devices, which are situated close to
end-users and data sources, thus minimizing constraints such as bandwidth consumption,
network congestion, response time, and operational costs imposed by cloud providers. To
enhance service provisioning agility and reduce infrastructure deployment costs, EC platforms frequently leverage technologies such as Network Functions Virtualization (NFV).
NFV facilitates the decoupling of hardware and network functions through virtualization
technologies, with network functions implemented as Virtual Network Functions (VNFs)
software entities. Network or even higher layers functions are implemented as Virtual
Network Functions (VNFs) software entities. The integration of EC and NFV paradigms,
as proposed by ETSI MEC, enables the creation of an ecosystem for 5G applications.
Such integration allows the creation of VNF chains, representing end-to-end services for
end-users and their deployment on edge nodes. A Service Function Chaining (SFC) comprises a set of VNFs chained together in a given order, where each VNF can be running
on a different edge node. The main challenges in this environment concern the dynamic
provisioning and deprovisioning of distributed resources to run the VNFs and meet application requirements while optimizing the cost to the infrastructure provider. In this sense,
scaling VNFs in this environment represents creating new containers or virtual machines
and reallocating resources to them due to the variation in the workload and dynamic nature of the EC environment. This work introduces an innovative approach for dynamically
scaling VNFs in the EC environment, employing a hybrid auto-scaling technique. This approach integrates an ensemble machine-learning technique that consists of different online machine-learning models to predict workload. It follows the MAPE-K (Monitor-AnalyzePlan-Execute over a shared Knowledge) control loop abstraction to dynamically adjust
resource allocation in response to workload changes. This approach is innovative because
it proactively predicts the workload to anticipate scaling actions and behaves reactively
when the prediction model does not meet the desired quality. In addition, the proposal
requires no prior knowledge of the data’s behavior, making it suitable for use in different
contexts. This work also proposes an algorithm to scale the VNF instances in the edge
computing environment that uses a strategy to define how many resources to allocate or
deallocate to a VNF instance during a scaling action. Finally, the ensemble method and
the proposed algorithm are evaluated, comparing prediction performance and the amount
of scaling actions and SLA violations. |
|---|