Suggestions on Instance Selection

Overview

Before purchasing an RDS for PostgreSQL instance, consider factors such as the price, performance, workload capacity, and workload scenario to choose the right one for your needs. This section describes the differences in instance types, billing modes, and storage types, helping you select the most suitable instance.

Storage Type

RDS for PostgreSQL provides cloud SSD (also called ultra-high I/O) and extreme SSD storage types. The table below describes their differences to help you select a storage type that best suits your workload needs.

**Table 1** Storage types
Storage Type	Characteristics	Use Case
Cloud SSD	Stores data in cloud disks for decoupled storage and compute. The maximum throughput is 350 MiB/s.	CPU-intensive sub-core business systems or application modules that need to minimize costs.
Extreme SSD	Uses 25GE network and RDMA technologies to provide you with up to 1,000 MiB/s throughput per disk and sub-millisecond latency.	Scenarios requiring high throughput. Core application systems that are sensitive to performance and have demanding requirements on storage I/O during peak hours, such as those in finance, e-commerce, government, and gaming.

Billing Mode

There are yearly/monthly and pay-per-use billing modes. Each one has different advantages and disadvantages.

**Table 2** Comparison of billing modes
Billing Mode	Yearly/Monthly	Pay-per-use
Payment	Prepaid Billed by the required duration specified in your order	Postpaid Billed for what you use
Supported Instance Types	Single-node Primary/Standby	Single-node Primary/Standby
Billing Period	Billed by the required duration specified in your order	Calculated by the second but billed every hour
Billed Items	Instance classes (vCPUs and memory), storage space, backup space, cross-region backup (optional), EIP (optional), and deployment in Dedicated Computing Cluster (optional)	Instance classes (vCPUs and memory), storage space, backup space, cross-region backup (optional), EIP (optional), and deployment in Dedicated Computing Cluster (optional)
Billing Mode Change	Yearly/Monthly billing can be changed to pay-per-use billing. The change is only applied after the yearly/monthly subscription expires. For details, see Yearly/Monthly to Pay-per-Use.	Pay-per-use billing can be changed to yearly/monthly billing. For details, see Pay-per-Use to Yearly/Monthly.
Scenarios	Recommended for resources expected to be in use long-term. This mode is less expensive than pay-per-use billing and is a good option for scenarios where the resource usage duration is predictable.	Good for short-term, bursty, or unpredictable workloads that cannot tolerate any interruptions, such as applications for e-commerce promotions, temporary testing, and scientific computing.

The table below lists the billed items of a yearly/monthly DB instance.

**Table 3** Billed items
Billed Item	Description	Billing Factor
* Instance class	vCPUs and memory	Billed by vCPUs, memory, and instance type. Computing and storage capabilities vary by the number of vCPUs and memory size.
* Storage space	Database storage	Billed based on unified standards.
* Backup space	RDS provides free backup space of the same size as your purchased database storage. After the free backup space is used up, charges are applied based on the backup space pricing details. Pricing is listed on a per-hour basis, but bills are calculated based on the actual usage duration.	Billed based on unified standards.
(Optional) Cross-region backup	RDS allows you to store backups in a region different from the one where your DB instance is located. Enabling cross-region backup will incur extra fees, and you will be billed for the storage space. Pricing is listed on a per-hour basis, but bills are calculated based on the actual usage duration.	Billed based on unified standards.
(Optional) EIP bandwidth	If an EIP is purchased along with a DB instance, the EIP is billed by bandwidth.	An EIP is required if a DB instance needs to access the Internet. Billed by bandwidth, traffic, and the EIP reservation price. EIP for a yearly/monthly DB instance: billed by bandwidth. EIP for a pay-per-use DB instance: billed by bandwidth, traffic, or shared bandwidth. You are also charged for IP reservation if you do not bind the EIP to any instance.
(Optional) Deployment in DCC	RDS can be deployed in DCC, which incurs extra fees.	Billed by the memory size.
(Optional) DRS migration	If you use Data Replication Service (DRS) for data migration, you will be billed based on the DRS pricing standard.	For details, see DRS Billing.

The table below lists the billed items of a pay-per-use DB instance.

**Table 4** Billed items
Billed Item	Description	Billing Factor
* Instance class	vCPUs and memory	Billed by vCPUs, memory, and instance type. Computing and storage capabilities vary by the number of vCPUs and memory size.
* Storage space	Database storage	Billed based on unified standards.
* Backup space	RDS provides free backup space of the same size as your purchased database storage. After the free backup space is used up, charges are applied based on the backup space pricing details. Pricing is listed on a per-hour basis, but bills are calculated based on the actual usage duration.	Billed based on unified standards.
(Optional) Cross-region backup	RDS allows you to store backups in a region different from the one where your DB instance is located. Enabling cross-region backup will incur extra fees, and you will be billed for the storage space. Pricing is listed on a per-hour basis, but bills are calculated based on the actual usage duration.	Billed based on unified standards.
(Optional) EIP bandwidth	If an EIP is purchased along with a DB instance, the EIP is billed by bandwidth.	An EIP is required if the DB instance needs to access the Internet. Billed by bandwidth, traffic, and the EIP reservation price. EIP for a yearly/monthly DB instance: billed by bandwidth. EIP for a pay-per-use DB instance: billed by bandwidth, traffic, or shared bandwidth. You are also charged for IP reservation if you do not bind the EIP to any instance.
(Optional) Deployment in DCC	RDS can be deployed in DCC, which incurs extra fees.	Billed by the memory size.
(Optional) DRS migration	If you use Data Replication Service (DRS) for data migration, you will be billed based on the DRS pricing standard.	For details, see DRS Billing.

DB Instance Type

RDS for PostgreSQL supports single-node and primary/standby instances. You can select an instance type tailored to your workload needs by referring to Table 5.

**Table 5** Instance types
DB Instance Type	Description	Advantages	Scenarios
Single-node	DB engine version: See DB Engines and Versions. Nodes: 1 primary node	Different from primary/standby instances that have two database nodes, a single-node instance has only one node, reducing the price to half of a primary/standby instance. If a fault occurs on a single-node instance, the instance cannot recover in a timely manner.	It is suitable for development and testing of microsites, and small- and medium-sized enterprises, or for learning about RDS.
Primary/Standby	DB engine version: See DB Engines and Versions. Nodes: 1 primary node + 1 standby node	The standby node of a primary/standby DB instance is only used for failover and restoration. It does not provide services. Since standby nodes cause extra performance overhead, the performance of single-node DB instances is similar to or even higher than primary/standby DB instances.	It is suitable for production databases of large- and medium-sized enterprises in Internet, Internet of Things (IoT), retail e-commerce sales, logistics, gaming, and other sectors.

Instance Class

Table 6 lists the instance classes of RDS for PostgreSQL instances billed on a pay-per-use or yearly/monthly basis.

**Table 6** Instance classes
Instance Class	Supported CPU Architecture	Description	Scenarios
General-purpose (recommended)	x86	CPU resources are shared with other general-purpose instances on the same physical machine. CPU usage is maximized through resource overcommitment. This instance class is a cost-effective option.	Suitable for scenarios where cost effectiveness is critical while performance stability is not critical
Dedicated (recommended)	x86	A dedicated instance has dedicated CPU and memory resources to ensure stable performance. There is no CPU resource contention between instances.	Suitable for core databases that require high stability, such as databases in e-commerce, gaming, finance, government, and enterprise
General-enhanced and general-enhanced II	x86	With a leading network acceleration engine and the fast packet processing mechanism of Data Plane Development Kit (DPDK), this instance class provides better network performance and more robust compute.	Suitable for websites and web applications that require high database compute and network performance

Functions Supported by Different Instance Types

**Table 7** Reference for instance type selection
Reference		Single-Node	Primary/Standby
Instance lifecycle	Stop an instance	√ Only pay-per-use instances can be stopped.	√ Only pay-per-use instances can be stopped.
	Start an instance	√ Only pay-per-use instances can be started.	√ Only pay-per-use instances can be started.
	Reboot an instance	√	√
	Delete a pay-per-use instance	√	√
	Recycle bin	√	√
Instance modification	Change a DB instance class	√	√
	Change the storage type	√	√
	Configure auto scaling of vCPUs and memory	x	√
	Manually scale storage space	√	√
	Configure storage autoscaling	√	√
	Configure the maintenance window	√	√
	Change a single-node instance to primary/standby	√	x
Version upgrade	Upgrade the kernel minor version	√	√
Version upgrade	Upgrade the major version on the console	x	√
Backup and restoration	Configure same-region backup policies	√	√
	Configure a cross-region backup policy	√	√
	Create a manual backup	√	√
	Download an instance-level backup	√	√
	Download an incremental backup	√	√
	Full restoration: restore data from backups	√	√
	Full restoration: PITR	√	√
	Database/table restoration: PITR	√	√
	Database-level restoration: restore data from backups	√	√
Read replica	Create read replicas	√	√
Read replica	Configure replication delay for a read replica	√	√
Data migration	Migration solution overview	√	√
Problem diagnosis and SQL analysis	Check the instance status	√	√
	Manage real-time sessions	√	√
	Create a SQL insights task	√	√
	Create a SQL throttling rule	√	√
Security and encryption	Reset the administrator password to restore root access	√	√
Security and encryption	Configure disk encryption	√	√
Parameter management	Modify instance parameters	√	√
Parameter management	Create a parameter template	√	√
Log management	View or download error logs	√	√
	View or download slow query logs	√	√
	Enable SQL audit	√	√
Metrics and alarms	View metrics	√	√
Metrics and alarms	Set alarm rules	√	√
Tags	Manage tags	√	√

Next topic: Using IAM to Grant Access to RDS

Feedback

Was this page helpful?

Helpful Not helpful

Provide feedback

Thank you very much for your feedback. We will continue working to improve the documentation.See the reply and handling status in My Cloud VOC.

The system is busy. Please try again later.

For any further questions, feel free to contact us through the chatbot.

Chatbot