MaxCompute:Install the Proxima CE package

Prerequisites

Complete Preparations before proceeding.

Download the package

Download proxima-ce-aliyun-1.0.2.jar.

The package contains the Proxima CE executable JAR file. After you upload it to your MaxCompute project as a resource, Proxima CE tasks can call it directly.

Upload the JAR file as a MaxCompute resource

Upload the JAR file to your MaxCompute project using either DataWorks or the MaxCompute client (odpscmd). Both methods register the file as a callable resource in your project.

Option 1: DataWorks

1. On the Data Development page of DataWorks, upload the JAR file as a JAR resource.

   Note

   For resources that are created or uploaded through the DataWorks visual interface:

   • If the resource has not been uploaded, you must select Upload to MaxCompute. If the resource has already been uploaded to MaxCompute, you must clear this checkbox. Otherwise, the upload fails.

   • If you select Upload to MaxCompute during the upload, the resource is stored in both DataWorks and MaxCompute. If you later delete the resource from MaxCompute by using a command, the resource in DataWorks still exists and can be viewed.

   • The resource name does not have to match the uploaded file name.

   

2. Commit and deploy the resource. Click the icon in the top toolbar of the resource configuration tab to commit the resource to the development environment.

   Note

   If a production task needs to use this resource, you must also deploy the resource to the production environment. For more information, see Deploy a task.

Option 2: odpscmd

Use the odpscmd command-line client to add the JAR file as a resource. See Add resources for the full command syntax.

Prepare input tables

Proxima CE requires two input tables for every vector search task:

• Doc table — the base table that contains all the vectors to search through.

• Query table — the table that contains the vectors you want to find nearest neighbors for.

Both tables share the same schema. Create them with the following SQL:

-- Create a doc table
CREATE TABLE doc_table_float_smoke(pk STRING, vector STRING <,category BIGINT>) PARTITIONED BY (pt STRING);

-- Create a query table
CREATE TABLE query_table_float_smoke(pk STRING, vector STRING <,category BIGINT>) PARTITIONED BY (pt STRING);

Output table format

After a vector search task completes, Proxima CE automatically writes results to an output table in MaxCompute — do not create it manually. Specify the table name using the -output_table parameter in the Proxima CE code.

What's next

With Proxima CE installed and your input tables ready, run your first vector search task. Choose a scenario based on your use case: