A video-streaming service hosted in the public cloud experiences short-lived traffic surges during live events but very low usage the rest of the week. Administrators want to meet these peaks without paying for idle capacity the rest of the time. Which provisioning method best satisfies both performance and cost goals?
Configure an auto-scaling group that adds or removes on-demand instances based on real-time metrics
Move the application to on-premises physical servers sized for the bursts
Spin up a fixed number of cloud VMs and leave them running behind a load balancer
Purchase three-year reserved instances sized for peak load
Autoscaling groups (or equivalent usage-based orchestration) watch metrics such as CPU or request count and automatically launch extra on-demand instances when thresholds are exceeded, then terminate them after the surge. Because billing stops when instances are shut down, the organization pays only for the capacity it actually uses. Buying long-term reserved capacity or keeping a fixed number of virtual machines online leaves resources idle and still incurs charges, while moving the workload to on-premises hardware involves capital expense and cannot be expanded quickly for unexpected bursts.
Ask Bash
Bash is our AI bot, trained to help you pass your exam. AI Generated Content may display inaccurate information, always double-check anything important.
What are temporary instances and how do they work in a usage-triggered system?
Open an interactive chat with Bash
What is autoscaling, and how does it contribute to cost savings?
Open an interactive chat with Bash
Why is shifting workloads to physical equipment less suitable for sudden traffic spikes?