Premium performance disks extend ESSD capabilities with I/O performance burst, Buffer Pool Extension (BPE), and data archiving to OSS for ApsaraDB RDS instances.
Features
A premium performance disk is a new storage type supported by ApsaraDB RDS. It is compatible with all features of enterprise SSDs (ESSDs) and supports I/O performance burst, Buffer Pool Extension (BPE), and data archiving to OSS.
-
Buffer Pool Extension (BPE)
BPE expands the buffer pool, improving read/write performance and reducing I/O jitter ondatabase operations. Use BPE to boost I/O without upgrading instance types. Buffer Pool Extension (BPE) feature.
-
I/O performance burst
I/O performance burst removes the IOPS cap during traffic spikes, delivering higher I/O without instance upgrades. Ideal for high, fluctuating I/O loads. I/O performance burst feature.
-
Data archiving
Data archiving moves infrequently accessed data to OSS, reducing storage costs by 80% compared to PL1 ESSDs. Data archiving feature.
Supported instances
RDS MySQL provides the High-availability Edition and the Cluster Edition.
-
Premium performance disks are supported for non-Serverless instances.
-
The serverless billing method is not supported.
Performance
Performance consists of baseline I/O and burst capacity.
|
Storage capacity of a premium performance disk |
Baseline performance |
Performance limit (Baseline + Burst) |
|
100 GB <= Storage capacity < 256 GB |
I/O bandwidth: IOPS: |
I/O bandwidth: IOPS: |
|
256 GB <= Storage capacity < 1000 GB |
I/O bandwidth: 4096 MB/s IOPS: |
|
|
Storage capacity >= 1000 GB |
I/O bandwidth: 4096 MB/s IOPS: 1,000,000 |
Performance of features
Buffer Pool Extension (BPE)
BPE improves read/write performance, shortensSQL execution time, and reduces I/O jitter impact on database operations. Buffer Pool Extension (BPE) feature.
I/O performance burst
Enabling I/O performance burst increases the maximum IOPS and throughput beyond baseline limits. The following table compares performance with and without burst.
Enabling I/O performance burst on the primary node also enables it on the secondary node. Read-only instances require separate enablement on each instance's details page.
|
Scenario |
Maximum IOPS |
Maximum throughput |
|
I/O performance burst disabled |
|
|
|
Burst enabled |
|
|
-
Throughput and IOPS are interdependent. Reaching the limit of one affects the other.
-
These formulas do not apply to general-purpose instance types, which share resources and cannot guarantee maximum IOPS or I/O bandwidth.
-
Maximum IOPS and I/O bandwidth per instance type: Instance types for standard primary ApsaraDB RDS for MySQL instances (original x86 architecture) and Instance types for Yitian primary ApsaraDB RDS for MySQL instances (original ARM architecture).
Data archiving
Archiving infrequently accessed data to OSS reduces storage costs. Access performance for archived data is lower than for non-archived data.
Scenarios
Buffer Pool Extension (BPE)
BPE suits workloads with high read I/O and frequent temporary table usage.
I/O performance burst
Best for workloads with high, fluctuating I/O demands.
Business peaks, large transactions, full table scans, and bulk queries can spike I/O. WhenCPU, memory, and storage are sufficient but I/O nears 100%, the only options are upgrading instance specs or storage level (for example, PL1 to PL2).
-
Without an upgrade, I/O exceeds the instance limit, causing performance degradation, query timeouts, and connection failures.
-
Upgrading instance specs resolves I/O bottlenecks but wastes CPU, memory, and storage.
-
Upgrading the storage level helps short-term, but wastes resources and increases costs after the I/O load drops.
I/O performance burst solves this by automatically raising the I/O limit during high load and restoring it afterward, providing serverless I/O performance and preventing wasted resources and unnecessary costs.
Data archiving
Data archiving suits instances with many infrequently accessed or modified tables.
Limits
-
Storage capacity ranges from 10 GB to 64,000 GB.
-
You cannot change a premium performance disk to another storage type. You can change an ESSD to a premium performance disk. Change an ESSD to a premium performance disk.
Billing
Billing methods
Premium performance disk fees = RDS storage space fees + I/O performance burst fees + Buffer Pool Extension (BPE) fees + Data archiving to OSS fees.
For the billing details of each feature, see the following topics:
BPE is free for ApsaraDB RDS for MySQL general-purpose instances.
No fees are charged for I/O performance burst before September 16, 2025 (Singapore time). Starting September 16, 2025, this feature will be billed. Official billing announcement.
RDS storage space fees
-
PL2 and PL3 ESSD fees differ from premium performance disk fees. The price difference shown on the console order at the time of change applies.
-
The shorter the high I/O duration, the more cost-effective a premium performance disk is compared to PL2 and PL3 ESSDs.
-
Premium performance disk fees = Storage space fees + I/O performance burst fees + Buffer Pool Extension (BPE) fees + Data archiving to OSS fees.
-
RDS storage space supports subscription and pay-as-you-go billing, priced the same as PL1 ESSDs. Billable items.
-
In the following regions, premium performance disks cost less than PL1 ESSDs regardless of billing method.
China (Hong Kong), Singapore, Malaysia (Kuala Lumpur), Indonesia (Jakarta), Philippines (Manila), Thailand (Bangkok), Japan (Tokyo), South Korea (Seoul), US (Silicon Valley), US (Virginia), Germany (Frankfurt), UK (London), and UAE (Dubai).
-
In the following regions, premium performance disks are priced the same as PL1 ESSDs.
China (Hangzhou), China (Shanghai), China (Beijing), China (Shenzhen), China (Nantong), China (Qingdao), China (Zhangjiakou), China (Hohhot), China (Ulanqab), China (Heyuan), China (Guangzhou), and China (Chengdu)