DataAI ETL Alteryx — Production Installation Guide

Installation and Usage

1. Verify prerequisites

For this 1.0.0 production candidate, use:

• Windows 64-bit;
• Alteryx Designer 2026.1;
• AMP Engine enabled;
• Java 17;
• a customer-managed Spark 3.5 / Scala 2.12 runtime;
• a working spark-submit command on the Designer machine or an absolute path to spark-submit.cmd, .bat, or .exe;
• a Spark catalog reachable with the identity that runs Designer.

Designer 2026.1 is the declared target because the YXI bundles Alteryx Python SDK 2.5.3 for embedded Python 3.13.11. A separately built and separately tested YXI is required for older Designer Python generations.

2. Verify the download

From PowerShell in the extracted package directory:

Get-FileHash .\install\DataAI_ETL_Alteryx_2026_1_Production_1.0.0.yxi -Algorithm SHA256

Compare the value with CHECKSUMS.sha256 from the provider package.

3. Confirm commercial entitlement

Read LICENSE.md and the executed Yanbor commercial agreement or order form. The tool requires the user to confirm that the organization has valid commercial rights before execution. The included commercial template is not an executed license.

4. Install the YXI

1. Close workflows that use older DataAI ETL tool versions.
2. Double-click DataAI_ETL_Alteryx_2026_1_Production_1.0.0.yxi, or drag it into Alteryx Designer.
3. Select user or administrator installation according to the customer's deployment policy.
4. Restart Designer if the DataAI ETL tool category does not appear.
5. Confirm DataAI ETL Quality appears in that category.

For Alteryx Server, an administrator must install the same approved YXI on every execution worker and make Java, Spark, catalogs, and connector libraries available to the service account. Validate Server separately before use.

5. Prepare a Spark source table

The tool operates on a Spark catalog table, not an in-memory Alteryx record stream. Load the sample CSV or an approved validation data set into a table, for example dataai_sample.customer_orders.

Ensure the Spark identity can:

• read the source table;
• create or overwrite only the selected output tables;
• write Delta tables in the selected catalog and schema.

6. Configure the tool

Create a new workflow and drag DataAI ETL Quality onto the canvas.

1. Spark Submit: enter the command or full path.
2. Spark Master: enter local[*], yarn, or the approved cluster URL.
3. Deploy Mode: select client or cluster.
4. Source Table: enter the existing Spark catalog table.
5. Enter any desired Clean, Rejected, Profile, and Findings tables. Each specified table is overwritten by the DataAI CLI.
6. Select normalization if desired.
7. Enter comma-separated record keys.
8. Paste a JSON array of rules, such as samples/rules.json.
9. Optionally enter a minimum quality score from 0 through 100.
10. Add extra Spark arguments as a JSON string array, never as one shell command. Example: ["--conf","spark.sql.session.timeZone=UTC"].
11. Set the timeout and confirm the organization's commercial entitlement.

The adapter enforces the DataAI main class and does not invoke a command shell. Do not place passwords, access tokens, or private keys in the tool fields. Use the customer's Spark credential and secret-management mechanisms.

7. Run and verify

1. Enable AMP for the workflow.
2. Connect a Browse tool to the optional Status output.
3. Run the workflow.
4. Confirm the status row says Succeeded and exit code 0.
5. Inspect the customer-controlled Spark logs.
6. Verify each configured output table and its row counts.
7. Test at least one intentionally failing rule and minimum-score gate.
8. Confirm no output table outside the selected schema changed.

The status output never returns customer rows or raw Spark logs. On failure, use local Designer and Spark logs, because logs can contain environment-specific details and are intentionally not copied into the Alteryx output.

8. Use advanced DataAI functions

The customer-delivery ZIP includes the API, quality, core, functions, and CLI JARs. Matrix balancing and the other advanced functions in docs/FUNCTION_CATALOG.md are called from a customer Java/Spark job. Add the four unshaded libraries to that job, create an approved wrapper for the desired function, and call that job from Alteryx or the customer's scheduler. They are not separate buttons in the first Alteryx tool version.

9. Uninstall

Use the customer's Alteryx custom-tool administration process to remove the DataAiEtlQuality_1_0 tool directory. Removing the tool does not delete Spark tables. Output tables must be retained or removed under the customer's data governance procedure.