ApsaraDB RDS:Primary ApsaraDB RDS instance types

Instance family

The instance family determines whether CPU, memory, storage, and I/O resources are dedicated to your instance or shared with other instances on the same physical server. This is the primary driver of performance consistency.

Choose an instance family based on your workload:

• Shared or General-purpose: Use for development environments, internal tools, or workloads where brief performance variation is acceptable. Because CPU resources are pooled across instances on the same physical server, the listed maximum connections and peak IOPS can be higher than those of a dedicated instance with the same vCPU and memory spec. However, peak performance is not guaranteed when the server is under high load.

• Dedicated: Use for production databases where consistent latency and throughput are required. All vCPU and memory resources are exclusively allocated to the instance, so the listed maximums are hard ceilings — not burst values.

• Dedicated host: Use when sharing any physical server resource with another tenant is unacceptable, such as for compliance-driven isolation or highly latency-sensitive workloads.

IOPS

Input/Output Operations Per Second (IOPS) measures how many random read and write requests the storage can handle per second.

The actual maximum IOPS of an instance is always the minimum of three independent ceilings: the instance type ceiling, the storage-capacity-derived ceiling, and the storage type ceiling. Use the formulas in the following table to calculate the actual maximum IOPS.

For instances using Premium Local SSDs, the actual maximum IOPS equals the maximum IOPS specified for the instance type — no formula applies.

For instances using cloud disks, apply the formula that matches the storage type:

The min{} function returns the smallest of the listed values. Adding storage capacity raises the capacity-derived ceiling but cannot push the result above the instance type ceiling or the storage type ceiling.

Example: IOPS calculation

Instance: mysql.x2.large.2c with a PL1 ESSD of 20 GB

Formula: min{Maximum IOPS for the instance type, 1800 + 50 × Storage capacity, 50000}

Result: 2800 IOPS — the smallest of the three values.

I/O bandwidth

I/O bandwidth (also called throughput) measures how much data the storage can read or write sequentially per second.

Like IOPS, the actual maximum I/O bandwidth is always the minimum of three independent ceilings: the instance type ceiling, the storage-capacity-derived ceiling, and the storage type ceiling.

For instances using cloud disks, apply the formula that matches the storage type:

Example: I/O bandwidth calculation

Instance: mysql.x2.large.2c with a PL3 ESSD of 5,000 GB

Formula: min{Maximum I/O bandwidth for the instance type, 120 + 0.5 × Storage capacity, 4000}

Result: 192 MB/s — the smallest of the three values.

Relationship between IOPS and I/O bandwidth

IOPS and I/O bandwidth are linked by the size of each I/O operation:

I/O bandwidth (MB/s) = IOPS × I/O size (KB) / 1024

An instance hits one limit before the other, depending on the workload:

• IOPS-bound workloads: Applications performing many small I/O operations (such as 4 KB random reads and writes) reach the IOPS ceiling first. I/O bandwidth utilization stays well below its limit.

• Bandwidth-bound workloads: Applications performing fewer but large I/O operations (such as 256 KB sequential reads and writes) saturate I/O bandwidth first. IOPS remains relatively low.