How Reserved Instances and Idle Billing Slash Serverless Costs by 30%

Technical Selection Scenario

Wang, a programmer managing on‑premise servers, faces high operational overhead and idle capacity costs. For two new online applications with strict latency, variable traffic, clear low‑traffic periods, and long startup times, he seeks a cloud solution that offers high elasticity, low cost, and minimal ops effort.

Three Architecture Options

Traditional : SLB + ECS + Auto Scaling requires manual load balancer setup, scaling policy tuning, and custom logging/monitoring, resulting in considerable deployment and maintenance effort.

Kubernetes : Offers Services, Ingress, and HPA for scaling, but HPA reacts on a minute‑scale, causing delayed scaling during traffic spikes; also demands extensive knowledge and ongoing ops for logging and monitoring.

Function Compute (FC) : Provides pre‑reserved mode with rapid auto‑scaling, built‑in HTTP triggers, comprehensive observability, and pay‑per‑use billing (no charge when idle), addressing all three of Wang’s requirements.

Reserved Mode and Cold‑Start Mitigation

FC supports both pay‑as‑you‑go and reserved modes. In reserved mode, a configured number of instances stay warm, eliminating cold‑start latency for long‑startup applications. When reserved instances are fully utilized, additional on‑demand instances are provisioned.

Idle Billing – The Cost‑Saving Weapon

Idle billing distinguishes between active and idle reserved instances, charging only 20% of the normal rate for idle CPU. This reduces costs by about 32% when utilization is 60% (e.g., 0.68 × original cost).

Enabling idle billing freezes CPU on idle instances, ensuring they only consume resources when handling requests.

Configuration Methods

Users can configure reserved instances and idle billing via the FC console’s Elastic Management page or through SDKs (Java, Go, Node.js) using the provided API.

Conclusion

Function Compute delivers a fully managed, high‑elasticity, low‑maintenance, and cost‑effective solution. The new idle‑billing feature further lowers expenses by charging only for truly used resources, reinforcing serverless as a powerful option for modern cloud workloads.