Skip to main content

metrics

Addedv0.1.x

This operation queries cluster performance metrics (QPS, latency, storage, etc.).

Description​

Zilliz Cloud organizes metrics into the following levels:

  • Organization-level metrics: Reflect account-wide status (e.g., license credits, usage) across all projects.

  • Cluster-level metrics: Reflect resource usage, performance, and data within individual clusters.

  • Collection-level metrics: A subset of cluster metrics broken down per collection, helping you pinpoint performance issues and plan capacity for individual collections.

Running this command without any options triggers a set of interactive prompts for you to set up the command.

Synopsis​

zilliz cluster metrics
--cluster-id <value>
--metric <value>
[--period <value>]
[--start <value>]
[--end <value>]
[--granularity <value>]
[--output <value>]

Options​

  • --cluster-id (string) -

    [REQUIRED]

    Indicates a cluster ID. For example, in01-xxxxxxxxxxxx.

    If a cluster is configured using zilliz context set, it automatically applies if this option is left unconfigured.

  • --metric, -m (array) -

    [REQUIRED]

    Indicates the metric name(s). You can chain this option after another to configure multiple metric names.

    Resource metrics:

    • CU_COMPUTATION

    • CU_CAPACITY

    • CU_SIZE

    • REPLICA_COUNT

    • STORAGE

    QPS metrics:

    • SEARCH_QPS

    • QUERY_QPS

    • INSERT_QPS

    • UPSERT_QPS

    • DELETE_QPS

    • BULK_INSERT_QPS

    Latency metrics:

    • SEARCH_LATENCY_AVG

    • SEARCH_LATENCY_P99

    • QUERY_LATENCY_AVG

    • QUERY_LATENCY_P99

    • INSERT_LATENCY_AVG

    • INSERT_LATENCY_P99

    • UPSERT_LATENCY_AVG

    • UPSERT_LATENCY_P99

    • DELETE_LATENCY_AVG

    • DELETE_LATENCY_P99

    VPS metrics:

    • SEARCH_VPS

    • INSERT_VPS

    • UPSERT_VPS

    • DELETE_VPS

    • BULK_INSERT_VPS

    Failure rate metrics:

    • SEARCH_FAIL_RATE

    • QUERY_FAIL_RATE

    • INSERT_FAIL_RATE

    • UPSERT_FAIL_RATE

    • DELTE_FAIL_RATE

    • BULK_INSERT_FAIL_RATE

    Data metrics:

    • ENTITIES

    • ENTITIES_LOADED

    • ENTITIES_INDEXED

    • COLLECTIONS

    • SLOW_QURIES

    Serverless metrics:

    • READ_VCU

    • WRITE_VCU

  • --period (string) -

    Indicates a relative time period from now.

    Uses d for days, m for months. The value defaults to 1h, indicating that the statistics are collected within the next hour.

  • --start (string) -

    Indicates the start time of a time range. For example, 2026-03-01 or 2026-03-01T10:00:00Z.

  • --end (string) -

    Indicates the end time of a time range. For example, 2026-03-15 or 2026-03-15T18:00:00Z.

  • --granularity, -g (string) -

    Indicates the data point interval. For example, 30s, 5m, 1h. This option defaults to auto.

  • --output, -o (string) -

    Indicates the output format. When this option is omitted, results are rendered as an in-terminal Braille chart visualization (since v1.3.1). Explicit values:

    • json,

    • table,

    • text.

Example​

zilliz cluster metrics -m READ_VCU -m WRITE_VCU