Modern systems often use automation for resource scaling. Current automatic scaling systems focus on systemgenerated metrics like CPU and memory utilization to determine scaling. Although these indicators are precise, they often lack insight into the operational performance of applications affected by scaling. Distributed traces model end-to-end requests or transactions as a hierarchical graph of spans, each representing a measured execution point. Trace data can reveal insights into latencies that are not captured by system-level metrics. To address this, we introduce SCALE, an automated resource scaler that uses distributed traces for scaling decisions. SCALE samples trace spans for duration anomalies and processes them through a heuristic-based modeling engine against defined latency thresholds. If thresholds are exceeded, SCALE triggers scaling commands for orchestration systems. SCALE is evaluated using trace data from Kubernetes microservice benchmarks.

Cloud systems are increasingly being managed by operation programs termed operators, which automate tedious, human-based operations. Operators of modern management platforms like Kubernetes, Twine, and ECS implement declarative interfaces based on the state-reconciliation principle. An operation declares a desired system state and the operator automatically reconciles the system to that declared state.