Journal of Information Processing
Online ISSN : 1882-6652
ISSN-L : 1882-6652
Performance Management of Cloud Populations via Cloud Probing
Marat Zhanikeev
Author information
JOURNAL FREE ACCESS

2016 Volume 24 Issue 1 Pages 99-108

Details
Abstract

Cloud population is a term that describes a cloud application distributed over many virtual machines or container-based boxes. Cloud platforms today offer simple tools for performance management (common example is load balancing) which are not sufficient for managing the performance of cloud populations. This paper proposes a new concept called cloud probing which is where applications themselves probe their host cloud platforms and optimize their own populations at runtime based on measurement data. This paper shows that even a simple optimization algorithm can lead to improved performance for the entire population. Since the only prerequisite function is the ability to migrate, the proposed method is also feasible in federated clouds where apps are fully in charge of managing their own populations spread across multiple cloud providers. This paper showcases the design of the TopoAPI that implements cloud probing, runs independently from physical platforms, and can therefore be used in federated environments.

Content from these authors
© 2016 by the Information Processing Society of Japan
Previous article Next article
feedback
Top