PolarDB Agent Memory offers a Standard Edition and an Enterprise Edition, designed for small and medium fully managed scenarios and large-scale enterprise scenarios that require reusing an existing PolarDB for MySQL cluster as vector storage, respectively. This topic describes how to create an Agent Memory.
Scope of application
-
Standard Edition: Suitable for small and medium scenarios. Supports up to 20 million vectors and 30 QPS. Uses shared compute resources in a fully managed mode for higher cost efficiency. Billed by the number of memory entries and LLM invocation fees.
-
Enterprise Edition: Currently in canary release. Suitable for large-scale, high-concurrency scenarios or scenarios that require reusing an existing PolarDB for MySQL cluster as vector storage. If you have related requirements, submit a ticket to contact us.
Procedure
Step 1: Go to the creation page
-
Log on to the PolarDB console.
-
In the left-side navigation pane, click PolarDB Agent Memory, and then click Create Memory.
Step 2: Select an edition
On the buy page, select Standard Edition or Enterprise Edition for Edition. The two editions differ in configuration items. Select an edition based on your business scale.
The edition cannot be changed after selection. Confirm your choice before creation.
Step 3: Configure common items (required for both editions)
Configure the following common items based on your requirements:
|
Parameter |
Description |
|
Memory Description |
Set a description for the PolarDB Agent Memory you are creating to facilitate identification and management later. Note
The description cannot start with http:// or https:// and must be 2 to 256 characters in length. |
|
Region |
Select the geographic location where the application resides. Note
|
|
Network Type |
Fixed to VPC. |
|
VPC |
Select the VPC where the PolarDB Agent Memory resides. We recommend that you select the same VPC as the ECS instances to which it connects. Otherwise, they cannot communicate over the internal network. Note
If existing VPCs do not meet your requirements, you can create a VPC and a vSwitch in the VPC console. |
|
vSwitch |
Select a vSwitch in the desired zone. |
|
SLR Role Authorization |
The system automatically detects the service-linked role AliyunServiceRoleForPolarDB. If the role has not been authorized, follow the console instructions to complete the authorization. |
|
Memory Type |
Only Plain-text Memory is supported. It is suitable for plain-text sessions and extracts text from sessions as memories. |
|
Memory Expiration Settings |
Select the validity period of memories:
|
|
Security Groups |
Select or create a security group to control access to the PolarDB Agent Memory. |
Step 4: Configure Enterprise Edition-specific items (required only for Enterprise Edition)
If you selected Enterprise Edition in Step 2, you also need to configure the following items:
|
Parameter |
Description |
|
Instance Class |
Select the compute specification for the Enterprise Edition PolarDB Agent Memory, for example, 2 vCPUs, 4 GB. The instance class determines the concurrent processing capability and memory extraction throughput. |
|
Database Association Mode |
Only Associate Existing Cluster is currently supported. The Enterprise Edition uses your existing PolarDB for MySQL cluster as the vector data storage backend. |
|
Select PolarDB Cluster |
Select a cluster from the PolarDB for MySQL cluster list in the current region to host the memory data. Note
We recommend that the PolarDB cluster and the Enterprise Edition PolarDB Agent Memory reside in the same VPC for optimal performance and security isolation. |
|
Database Name |
The name of the database used to store memory data. This field is required. Note
If the target database does not exist, you can create it in advance on the page of the selected PolarDB cluster. |
|
The username used to log on to the database. |
The account used to access the database specified in the previous step. You can select an existing Privileged Account in the selected PolarDB cluster, or a standard account that has read and write permissions on the specified database. |
|
Account Password and Confirm password |
Enter the password for the database account specified above, and enter it again for confirmation. |
Step 5: Purchase and view
-
Confirm that the Configuration Summary on the right side is correct, click Buy Now, and complete the payment.
-
After the purchase is successful, return to the PolarDB console. You can view the newly created PolarDB Agent Memory on the PolarDB Agent Memory list page.
NoteThe system requires 3 to 5 minutes to complete application initialization. Please wait patiently.
What to do next
After creation, you can obtain the endpoint, configure the whitelist, and start using the PolarDB Agent Memory. For more information, see Use PolarDB Agent Memory.