Skip to content
By using AWS re:Post, you agree to the AWS re:Post Terms of Use
AWS re:Postre:Post
About re:Post

Issues with AWS Supply Chain Demand Planning Module: Long Execution Time and Internal Error

0

We're working on a Proof of Concept that explores the capabilities of the Demand Planning module in AWS Supply Chain.

We've closely followed the official documentation and ensured that all required datasets were uploaded and properly formatted. However, we're encountering two major issues:

  1. Excessive processing time – the demand planning process can take up to three days, even when running with a very small sample dataset.
  2. Internal error – after processing, the job fails with a generic internal error message, and no detailed logs or diagnostics are provided, which makes troubleshooting extremely difficult.

This behavior is currently blocking our progress.

Additionally, we're trying to understand whether AWS Supply Chain internally uses Amazon Forecast. Some sources suggest this might be the case, but the documentation doesn't clearly confirm it. This would be relevant to our architecture decisions, especially considering that newer AWS accounts no longer have direct access to Forecast.

Any guidance or insight would be greatly appreciated.

Topics
Management & GovernanceMachine Learning & AIBusiness Applications
Tags
AWS Supply ChainAmazon ForecastAWS Management Console
Language
English

asked a year ago138 views

1 Answer
0

I understand you're facing significant challenges with your AWS Supply Chain Demand Planning proof of concept, specifically excessive processing times (up to three days) and generic internal errors, even with a small dataset. Let me help you troubleshoot these issues and answer your question about the relationship with Amazon Forecast. Common Causes and Solutions for Your Issues

  1. Excessive Processing Time (3+ Days)

This is unusually long, especially for a small dataset. Here are potential causes and solutions: Data Structure and Quality Issues

Inconsistent Time Series: Ensure your historical data has consistent time intervals without gaps
Solution: Validate that your date formats are consistent and that you have continuous data points

Configuration Problems

Forecast Horizon Too Long: Setting an extensive forecast period increases computation time
Solution: Start with a shorter forecast horizon (e.g., 4-8 weeks) for your POC

Resource Constraints

Instance Sizing: AWS Supply Chain might be allocating limited resources for your instance
Solution: Contact AWS Support to check if your instance needs resource adjustment

2. Internal Error with No Detailed Logs

This is particularly frustrating for troubleshooting. Here's what you can try: Data Validation

Required Fields: Double-check that all mandatory fields are present and properly formatted
Reference Data: Ensure all reference data (like product hierarchies) is complete and consistent

Common Error Triggers

Special Characters: Remove any special characters from your dataset
Null Values: Handle null values appropriately (either fill or remove)
Date Range Issues: Ensure your historical data covers at least 2-3x the forecast horizon

Step-by-Step Troubleshooting Approach

Start with Minimal Dataset
    Create an extremely simplified dataset with just 1-2 products and clean historical data
    Test the process with this minimal dataset to isolate data-related issues

Check CloudTrail Logs
    AWS Supply Chain operations are logged in CloudTrail
    Look for errors related to your AWS Supply Chain instance

Review Data Connections
    Verify that all data connections are properly established
    Check if any data ingestion errors are occurring

Examine Data Lake Structure
    AWS Supply Chain uses a data lake architecture
    Ensure your S3 buckets and permissions are correctly configured

Contact AWS Support
    Given the severity of the issues (3-day processing and internal errors)
    Request escalation to the AWS Supply Chain specialist team
    Ask specifically about resource allocation for your instance

AWS Supply Chain and Amazon Forecast Relationship

To answer your specific question: Yes, AWS Supply Chain Demand Planning does use Amazon Forecast internally, but with some important distinctions:

AWS Supply Chain uses a specialized implementation of Amazon Forecast that's optimized for supply chain use cases
It includes pre-built integrations for supply chain data models and specialized algorithms
The service abstracts away much of the complexity of directly using Amazon Forecast
It adds supply chain-specific features like product hierarchies and attribute handling

This architecture explains some of the behavior you're seeing:

The processing time includes data preparation, model training, and forecast generation steps from Amazon Forecast
The "black box" nature of the errors is due to the abstraction layer between you and the underlying Forecast service

Recommended Next Steps

Simplify Your Dataset
    Reduce to absolute minimum required fields
    Start with just a few products and a short time horizon

Check Data Format Requirements
    Review the AWS Supply Chain data model documentation 

    Ensure strict adherence to required formats

Engage AWS Support
    Open a support case specifically mentioning:
        The 3-day processing time for a small dataset
        The internal error with no diagnostics
        That this is blocking your POC progress

Consider Direct Amazon Forecast
    For your POC, you might consider testing Amazon Forecast directly
    This could provide more visibility into the forecasting process and errors

Request Enhanced Logging
    Ask AWS Support if enhanced logging can be enabled for your instance
    This might provide more detailed error information

The issues you're experiencing are not typical, even for a new service like AWS Supply Chain. With a properly formatted small dataset, you should expect processing times of hours, not days. The lack of detailed error messages is also something that AWS Support should be able to help address.

https://docs.aws.amazon.com/aws-supply-chain/latest/userguide/demand-planning.html

https://docs.aws.amazon.com/aws-supply-chain/latest/userguide/Troubleshooting.html

https://docs.aws.amazon.com/aws-supply-chain/latest/userguide/transactional.html

https://docs.aws.amazon.com/aws-supply-chain/latest/userguide/forecast-algorithims.html

AWS

answered a year ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Relevant content