New Observability Updates in Confluent Cloud Kafka

Confluent Cloud has shipped several new Kafka observability features via its Metrics API. Key additions include: a `client_limit_milliseconds` metric that identifies which principals are being throttled and why, collapsing multi-hour debugging loops; a `max_pending_rebalance_time_milliseconds` metric for directly tracking consumer group rebalance frequency and duration; a new `connection_accept_count` metric to correlate connection storms with elastic cluster scaling; and an upcoming `cleanup.policy` dimension on `partition_count` to differentiate compacted partitions. The Confluent Cloud Console also gets a refreshed Cluster Monitoring page with an eCKU usage chart showing which dimensions drive capacity consumption. Upcoming work includes throttled client visibility in the console and consumer group lag support for empty groups.