I would like to explore all performance metrics to evaluate an auto-scaling system.
The desired auto-scaling system scales out/in a web application, which is hosted in the cloud environment.
I know some metrics such as (1) Cost, (2) Response time, (3) Delay time, (4) Resource utilization, (5) Total Provisioned VMs, (6) Oscillation mitigation (or scaling overload), (7) Time to Adaptation, (8) VM minute, (9) Contradictory scaling decisions, (10) Contradictory scaling actions, (11)The rate of SLA Violation, (12) Stability, . . .
anymore? If yes, you can be a part of our team.
I have also used the following for vertical elasticity:
- Efficiency: the ratio of utilised capacity to allocated capacity
- Variability: amount of variability in performance/QoS metric e.g. throughput or response time derived using control theoretic measures such as Integral of Absolute Error
- Adaptation: (which you have listed), you could also break it done into time to adaptation after a scale up or after a scale up
Thank you, Dr. Ilbidunmoye,
I think efficiency (= (4) Resource Utilization) has been listed.
Can you explain Variability?
We collected all metrics
in this article:
Article Performance Evaluation Metrics for Cloud, Fog and Edge Compu...