Cloud infrastructures can provide resource sharing between many applications and usually can meet the requirements of most of them. However, in order to enable an efficient usage of these resources, automatic orchestration is required. Commonly, automatic orchestration tools are based on the observability of the infrastructure itself, but that is not enough in some cases. Certain classes of applications have specific requirements that are difficult to meet, such as low latency, high bandwidth and high computational power. To properly meet these requirements, orchestration must be based on multilevel observability, which means collecting data from both the application and the infrastructure levels. Thus in this work we developed a platform aiming to show how multilevel observability can be implemented and how it can be used to improve automatic orchestration in cloud environments. As a case study, an application of computer vision and robotics, with very demanding requirements, was used to perform two experiments and illustrate the issues addressed in this paper. The results confirm that cloud orchestration can largely benefit from multilevel observability by allowing specific application requirements to be met, as well as improving the allocation of infrastructure resources.
|Title of host publication||2018 IEEE 16th International Conference on Dependable, Autonomic and Secure Computing, 16th International Conference on Pervasive Intelligence and Computing, 4th International Conference on Big Data Intelligence and Computing and 3rd Cyber Science and Technology Congress (DASC-PICom-DataCom-CyberSciTech 2018)|
|Subtitle of host publication||Proceedings of a meeting held 12-15 August 2018, Athens, Greece|
|Publisher||Institute of Electrical and Electronics Engineers (IEEE)|
|Number of pages||6|
|Publication status||Published - Feb 2019|
- Intelligent spaces
- cloud computing