Cloud-Internet of Things computing paradigms call for novel and efficient methodologies to decide where to place application services in continuity with Continuous Integration/Continuous Deployment pipelines and infrastructure monitoring. In this article, we present Continuous Adaptation (CA), a new DevOps practice for (1) detecting runtime changes in the application requirements or the infrastructure that, due to their change in resource consumption or their effects on the Quality of Service (QoS), can affect the validity and dependability of the current application placement, and for (2) locally intervening on them by suggesting new placements that ensure all (functional and non-functional) application requirements are met. We assess a prototype of CA, ConDADO, and analyze its performance over a motivating use case. ConDADO adapts the application placement to environmental changes through the use of continuous reasoning, reducing the size of the problem to be solved to optimize its performance. The evaluation shows that ConDADO is able to obtain nearly optimal QoS up to 4.2× 4.2× faster than alternative techniques, also minimizing the cost of service migration.
Continuous QoS-aware adaptation of Cloud-IoT application placements
Herrera, Juan Luis
;Forti, Stefano;Brogi, Antonio;
2023-01-01
Abstract
Cloud-Internet of Things computing paradigms call for novel and efficient methodologies to decide where to place application services in continuity with Continuous Integration/Continuous Deployment pipelines and infrastructure monitoring. In this article, we present Continuous Adaptation (CA), a new DevOps practice for (1) detecting runtime changes in the application requirements or the infrastructure that, due to their change in resource consumption or their effects on the Quality of Service (QoS), can affect the validity and dependability of the current application placement, and for (2) locally intervening on them by suggesting new placements that ensure all (functional and non-functional) application requirements are met. We assess a prototype of CA, ConDADO, and analyze its performance over a motivating use case. ConDADO adapts the application placement to environmental changes through the use of continuous reasoning, reducing the size of the problem to be solved to optimize its performance. The evaluation shows that ConDADO is able to obtain nearly optimal QoS up to 4.2× 4.2× faster than alternative techniques, also minimizing the cost of service migration.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
