Cloud Console metrics

The Aerospike Cloud Console displays a subset of metrics for observing your cluster at a glance. These metrics are derived from Aerospike Database Prometheus metrics collected by the Aerospike Prometheus Exporter running alongside each cluster node and served through the Cloud Public API.

Caution

Cloud Console metrics are intended for quick health checks only. Do not use them for production monitoring or alerting. For comprehensive monitoring, use the Aerospike Database metrics endpoint instead.

These metrics come directly from the public API, see Public API Metrics for more information.

Usage

Check the usage gauges to understand how much data storage and memory your cluster is consuming. Use these gauges to determine when to scale your cluster or adjust your data model.

• Data shows the percentage of storage remaining for writes. As this gauge approaches zero, scale your cluster or reduce stored data to avoid write failures. The total data capacity is data_total_bytes divided by the replication factor, and the gauge value reflects data_avail_pct.

• Memory shows processing memory consumption across the cluster. Review this gauge to verify your cluster has enough headroom for operations. The total memory capacity is the instance type memory multiplied by the cluster size, and the gauge value reflects the average of system_free_mem_pct across all nodes.

Latency

Check latency to understand how quickly your cluster responds to client requests. The Aerospike Cloud Console displays latency at the p95, p99, and p99.9 quantiles for read and write operations, averaged over the past day. Rising latency at higher quantiles can indicate resource contention, hot keys, or the need to scale.

These quantiles are calculated using the Prometheus histogram_quantile function on the average of read_latency_avg and latency_ms.

Transactions per second (TPS)

Check transactions per second (TPS) to understand the throughput of read and write operations over time. The Aerospike Cloud Console displays TPS as a line graph with an adjustable time range. Use TPS to identify traffic patterns, spot unexpected spikes or drops, and correlate throughput changes with latency or usage shifts.

TPS is calculated using the Prometheus increase function on the following namespace-level metrics:

• Read TPS: Sum of the increase of client_read_error, client_read_not_found, client_read_success, and client_read_timeout.
• Write TPS: Sum of the increase of client_write_error, client_write_success, and client_write_timeout.