ApsaraDB RDS for MySQL provides the following storage types: local SSDs, premium performance disks, enhanced SSDs (ESSDs), and standard SSDs (some RDS instances are no longer available for purchase). This topic describes the differences among these storage types and provides purchase suggestions.
Storage type description
Storage type | Description |
Local SSD | Local SSDs are SSD disks that are located on the same node as the database engine and provide low I/O latency. |
Premium performance disk | Premium performance disks are compatible with all features of ESSDs. Premium performance disks provide better baseline I/O performance than PL1 ESSDs and offer the following four features:
|
ESSD | ESSDs are developed by Alibaba Cloud and deliver ultra high storage performance. ESSDs are based on the next-generation distributed block storage architecture. ESSDs are integrated with 25 Gigabit Ethernet and remote direct memory access (RDMA) technologies. ESSDs can provide lower one-way latencies than standard SSDs and process up to 1 million read and write requests at random per second. ESSDs are provided in the following performance levels (PLs):
For more information about the performance of ESSDs, see ESSD. |
Standard SSD | Standard SSDs are elastic block storage devices that are designed based on a distributed storage architecture. You can store data in standard SSDs to separate computing from storage. Note Standard SSDs are being phased out. We recommend that you use ESSDs. For more information, see [Discontinued/End of Support] Standard SSDs are no longer available for some RDS instances from July 1, 2022. |
For a comparison of performance metrics (single disk capacity, maximum IOPS, and maximum throughput) among these storage types, see Block storage performance.
All of the preceding storage types meet the reliability, durability, and read and write performance goals that are specified in the Alibaba Cloud service level agreement (SLA).
High-performance local disk: The associated RDS instances all have a primary-secondary (high availability) architecture. When the primary node fails, the switchover between the primary and secondary nodes completes within seconds.
Cloud disks (standard SSDs, ESSDs, and premium performance disks): Cloud disks are distributed cloud disks that ensure data reliability through multiple replicas. If you use RDS High-availability Edition or RDS Cluster Edition and the primary RDS instance fails, a primary/secondary switchover is performed within seconds.
View the storage type
You can view the storage type of an RDS instance on the Basic Information page of the instance.
Comparison among storage types
Currently, only RDS MySQL instances that run RDS High-availability Edition support local SSDs.
Metric | Local SSD | Premium performance disk | ESSD |
I/O performance | ★★★★★ Delivers low I/O latency and high I/O performance:
| ★★★★★★ Provides better baseline I/O performance than PL1 ESSDs and offers the Buffer Pool Extension (BPE) feature, I/O burst feature, and data archiving feature. The following list describes the I/O performance:
| ★★★★★ Delivers higher I/O performance than standard SSDs:
|
Configuration flexibility | ★★★★ Provides various configuration options and allows you to separately adjust the storage capacity. For some RDS instances that use local SSDs, the storage capacity is bound to the instance type and cannot be separately adjusted. | ★★★★★ Provides various configuration options and allows you to expand or reduce the storage capacity of an RDS instance. Note Only some MySQL instances that meet specific requirements support storage capacity reduction. For more information, see Overview of instance changes and Change the specifications of an ApsaraDB RDS instance. | ★★★★★ Provides various configuration options and allows you to expand or reduce the storage capacity of an RDS instance. Note Only some MySQL instances that meet specific requirements support storage capacity reduction. For more information, see Overview of instance changes and Change the specifications of an ApsaraDB RDS instance. |
Backup method | Provides physical backups by using XtraBackup | Provides snapshot backups | Provides snapshot backups |
Time required for backup, read-only instance creation, and instance cloning | ★★ Requires a few hours based on the disk space. | ★★★★★ Requires a few seconds. | ★★★★★ Requires a few seconds. |
Duration | ★★★ Requires a few hours to copy data when you scale up the storage. | ★★★★★ Supports online storage capacity expansion. You can expand the storage capacity of an RDS instance within a few seconds. | ★★★★★ Supports online storage capacity expansion. You can expand the storage capacity of an RDS instance within a few seconds. |
Scale-out impact | Transient connections occur during the scale-out process. | None. | None. |
Data durability | ★★★★ Has a risk of data corruption due to hardware faults, which requires secondary databases to support disaster recovery. | ★★★★★ Provides 99.9999999% data reliability and supports RDS instances that run RDS Basic Edition to reduce costs. | ★★★★★ Provides 99.9999999% data reliability and supports RDS instances that run RDS Basic Edition to reduce costs. |
Purchase suggestions
We recommend that you prioritize premium performance disks:
If your I/O volume is large, we recommend that you enable the Buffer Pool Extension (BPE) feature.
If your I/O fluctuates significantly, we recommend that you enable the I/O burst feature.
If your database contains many infrequently accessed or modified tables, we recommend that you enable the data archiving feature.
If cloud disks of specific PLs are required, you can purchase ESSDs.
Supported instance types, engine versions, and architectures
For more information about the storage types and features that are supported by each instance type, see Features of ApsaraDB RDS for MySQL.