Getting to the Bottom of Serverless Billing

2025-06-02

AI Summary

This paper reveals significant cost inflation in public cloud serverless platforms, where billing mechanisms—including wall-clock-time pricing, fixed invocation fees, and resource allocation rounding—cause billed resources to exceed actual consumption by up to 5.49×. To systematically characterize this issue, we adopt a top-down, multi-layered empirical methodology integrating micro-benchmarks, production traffic modeling, and kernel-level scheduler tracing. This is the first end-to-end analysis of the billing pipelines across AWS Lambda, Azure Functions, and Google Cloud Functions. Our investigation uncovers novel, previously undocumented cost drivers: service architecture–induced execution patterns, idle-period resource over-provisioning due to keep-alive semantics, and misalignment between OS scheduling granularity and serverless resource allocation units. These findings yield actionable optimization strategies and design principles for building cost-efficient serverless systems.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Machine Learning: Hardware-aware ML
• Search and Optimization: Distributed Search

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Systems and Infrastructure for Web, Mobile and WoT: Virtualization and resource management in Web systems and infrastructures
• User Modeling, Personalization and Recommendation: Practical large-scale studies of user experience

Abstract

Public cloud serverless platforms have attracted a large user base due to their high scalability, plug-and-play deployment model, and pay-per-use billing. However, compared to virtual machines and container hosting services, modern serverless offerings typically impose higher per-unit time and resource charges. Additionally, billing practices such as wall-clock allocation-based billing, invocation fees, and usage rounding up can further increase costs. This work, for the first time, holistically demystifies these costs by conducting an in-depth, top-down characterization and analysis from user-facing billing models, through request serving architectures, and down to operating system scheduling on major public serverless platforms. We quantify, for the first time, how current billing practices inflate billable resources up to 5.49x beyond actual consumption. Also, our analysis reveals previously unreported cost drivers, such as operational patterns of serving architectures that create overheads, details of resource allocation during keep-alive periods, and OS scheduling granularity effects that directly impact both performance and billing. By tracing the sources of costs from billing models down to OS scheduling, we uncover the rationale behind today's expensive serverless billing model and practices and provide insights for designing performant and cost-effective serverless systems.

Problem

• Analyzes serverless billing costs and hidden cost drivers
• Quantifies billing inflation up to 5.49x actual resource use
• Explores OS scheduling impacts on performance and billing

Innovation

• Top-down analysis of serverless billing models
• Quantifies billing inflation up to 5.49x
• Reveals hidden cost drivers in OS scheduling