INACCURATE MODELLING OF RESOURCE USAGE

Cloud services encounter distinctive challenges in resource management and provisioning. One critical vulnerability is resource exhaustion, stemming from various factors intrinsic to the ever-changing nature of cloud environments. Here is a breakdown of these challenges and their potential causes:

Statistical Provisioning

Cloud resources are usually provisioned statistically, where providers allocate resources based on predicted demand rather than specific customer needs. While this method offers flexibility and scalability, it can lead to overbooking or over-provisioning if resource usage modeling is inaccurate, resulting in wasted resources.

Resource Allocation Algorithms

Cloud providers utilize resource allocation algorithms such as Token Bucket, Fair Queuing, and Class Based Queuing to efficiently manage resource allocation. However, these algorithms may be susceptible to fairness distortions, causing imbalances in resource distribution and potentially favoring certain customers or applications over others, affecting overall service quality.

Extraordinary Events

Unforeseen events like sudden spikes in demand due to external news events or content delivery surges can strain resource allocation algorithms. These events, not adequately factored into the resource provisioning model, can lead to performance degradation or service interruptions.