Leveraging Core and Uncore Frequency Scaling for Power-Efficient Serverless Workflows

TL;DR

Ωkypous tackles the challenge of power-efficient serverless workflows under end-to-end SLOs by jointly optimizing Core and Uncore frequencies using a lightweight global grey-box latency predictor and a slack-aware adaptive controller. The framework budgets latency at the function level, dynamically re-allocates remaining slack, and selects the lowest-power CFS/UFS configuration that satisfies the current budget, while a conflict resolver handles co-location effects. Key contributions include a two-phase latency budgeting approach (pre-deployment CP-based budgeting and runtime dynamic updates), a global PMC-based latency model with few-shot generalization achieving ≈$MAPE \approx 4\%$, and robust evaluation against Linux governors and state-of-the-art DVFS schemes on Azure traces, showing ≈$16\%$ power savings and SLO violations ≈$1.8\%$ under power caps. The results demonstrate practical, per-invocation DVFS decisions that leverage Uncore frequency to unlock additional power headroom, offering tangible data-center energy benefits for serverless platforms.

Abstract

Serverless workflows have emerged in Function-as-a-Service (FaaS) platforms to represent the operational structure of traditional applications. With latency propagation effects becoming increasingly prominent, step-wise resource tuning is required to address Service-Level-Objectives (SLOs). Modern processors' allowance for fine-grained Dynamic Voltage and Frequency Scaling (DVFS), coupled with serverless workflows' intermittent nature, presents a unique opportunity to reduce power while meeting SLOs. We introduce $Ω$kypous, an SLO-driven DVFS framework for serverless workflows. $Ω$kypous employs a grey-box model that predicts functions' execution latency and power under different Core and Uncore frequency combinations. Based on these predictions and the timing slacks between workflow functions, $Ω$kypous uses a closed-loop control mechanism to dynamically adjust Core and Uncore frequencies, thus minimizing power consumption without compromising predefined end-to-end latency constraints. Our evaluation on real-world traces from Azure, against state-of-the-art power management frameworks, demonstrates an average power consumption reduction of 16\%, while consistently maintaining low SLO violation rates (1.8\%), when operating under power caps.

Figures (17)

• Figure 1: Impact of Core and Uncore frequency scaling on tail latency and power consumption.
• Figure 2: Pareto analysis of latency and power consumption trade-offs, between i) CFS&UFS (beige), ii) CFS only (blue) and iii) CPU Quotas (orange).
• Figure 3: Performance variability across function executions in serverless workflows.
• Figure 4: Latency modeling challenges.
• Figure 5: $\boldsymbol{\Omega}$kypous optimization approach overview.