All Products
Search

This Product

    ApsaraDB RDS:[Product changes/Feature changes] I/O performance-related enhanced metrics are supported and calculation methods of the metrics are optimized in ApsaraDB RDS for PostgreSQL

    Document Center

    ApsaraDB RDS:[Product changes/Feature changes] I/O performance-related enhanced metrics are supported and calculation methods of the metrics are optimized in ApsaraDB RDS for PostgreSQL

    Last Updated:Apr 11, 2025

    Starting March 20, 2025, I/O performance-related enhanced metrics are supported to improve the accuracy of IOPS and I/O bandwidth monitoring for ApsaraDB RDS for PostgreSQL instances that use dedicated instance types. This helps users better understand the storage performance of ApsaraDB RDS for PostgreSQL instances.

    Effective date

    From March 20, 2025 to March 24, 2025

    Description

    I/O performance-related metrics are improved, and calculation methods are optimized. The following table describes the newly supported metrics and calculation methods.

    New metrics and calculation methods

    Metric

    Description

    Storage type

    Calculation method

    os.iops.limit

    Baseline IOPS limit

    General Enterprise SSD (ESSD)

    min{Maximum IOPS for the instance type, IOPS increment + max[min(1800 + 50 × Disk size in GB, 50000), 3000]}

    PL0 ESSD

    min{Maximum IOPS for the instance type, min((1800 + 12 × Disk size in GB), 10000)}

    PL1 ESSD

    min{Maximum IOPS for the instance type, min((1800 + 50 × Disk size in GB), 50000)}

    PL2 ESSD

    min{Maximum IOPS for the instance type, min((1800 + 50 × Disk size in GB), 100000)}

    PL3 ESSD

    min{Maximum IOPS for the instance type, min((1800 + 50 × Disk size in GB), 1000000)}

    Standard SSD

    min{Maximum IOPS for the instance type, min((1800 + 30 × Disk size in GB), 25000)}

    os.iops.burst_limit

    IOPS burst limit

    Note

    This metric is suitable for RDS instances for which the I/O burst feature of general ESSDs is enabled.

    General ESSD

    min(Maximum IOPS for the instance type, 1000000, 1000 × Disk size in GB)

    os.iothroughput.limit

    Baseline I/O bandwidth

    General ESSD

    min{Maximum I/O bandwidth for the instance type, 16 × IOPS increment/1024 + max[min(120 + Disk size in GB/2, 350), 125]}

    PL0 ESSD

    min{Maximum I/O bandwidth for the instance type, (100 + 0.25 × Disk size in GB), 180}

    PL1 ESSD

    min{Maximum I/O bandwidth for the instance type, (120 + 0.5 × Disk size in GB), 350}

    PL2 ESSD

    min{Maximum I/O bandwidth for the instance type, (120 + 0.5 × Disk size in GB), 750}

    PL3 ESSD

    min{Maximum I/O bandwidth for the instance type, (120 + 0.5 × Disk size in GB), 4000}

    Standard SSD

    min{Maximum I/O bandwidth for the instance type, (120 + 0.5 × Disk size in GB), 300}

    os.iothroughput.burst_limit

    Burst I/O bandwidth limit

    Note

    This metric is suitable for RDS instances for which the I/O burst feature of general ESSDs is enabled.

    General ESSD

    min{Maximum I/O bandwidth for the instance type, min[min(1000000, 1000 × Disk size in GB) × 16/1024, 4096]}

    os.io_usage.iops_usage

    IOPS usage

    General ESSD, ESSD, and standard SSD

    os.iops.total/os.iops.limit

    os.io_usage.iops_burst_ratio

    Burst IOPS rate

    General ESSD

    os.iops.burst_limit/os.iops.limit

    os.io_usage.mbps_usage

    I/O bandwidth usage

    General ESSD, ESSD, and standard SSD

    os.iothroughput.total/os.iothroughput.limit

    os.io_usage.mbps_burst_ratio

    Burst I/O bandwidth rate

    General ESSD

    os.iothroughput.burst_limit/os.iothroughput.limit

    Impacts

    • ApsaraDB RDS synchronizes the updates of the calculation methods for I/O performance-related metrics to the existing RDS instances that use dedicated instance types. During optimization, the preceding metrics may temporarily fluctuate. However, this adjustment only involves the calculation methods of monitoring data and does not affect the actual I/O performance of RDS instances.

    • The host on which the Elastic Compute Service (ECS) instance resides has burst bandwidth. As a result, the ECS instance may encounter burst bandwidth, and the actual bandwidth usage may exceed the baseline bandwidth limit. This is a normal case. A service level agreement (SLA) is not guaranteed for the burst bandwidth, and you are not charged additional fees.

    • The maximum IOPS and maximum I/O bandwidth cannot be guaranteed for RDS instances that use general-purpose instance types. If your workloads are sensitive to IOPS, we recommend that you use dedicated instance types.

    References