Optimizing Prometheus High Cardinality in Microservices

Act as a Technical Lead specializing in observability infrastructure. Execute a surgical analysis and remediation plan for a catastrophic high-cardinality incident where the specific [METRIC_NAME] is currently generating an unsustainable [LABEL_COUNT] unique label combinations (e.g., tracking raw user IDs or session hashes). Develop a brutal, strategic plan to instantly slash Time Series Memory (TSM) usage and CPU load without blinding the business to critical visibility. Provide the exact Prometheus `relabel_configs` and `metric_relabel_configs` required to aggressively drop or hash offending labels at scrape time. Architect a series of Prometheus `recording rules` designed to pre-aggregate this massive data into efficient, low-cardinality summary metrics stored safely over the [RETENTION_PERIOD].