aws-samples / amazon-textract-idp-cdk-constructs

MIT No Attribution
30 stars 13 forks source link

Allow for millions of document processing #102

Closed schadem closed 1 year ago

schadem commented 1 year ago

At the moment the processing relies on retries, which

The solution is to only allow the number of workflows at any given time close to the concurrenty for the component with the lowest limit (aka bottleneck). This could be a backend system that only allows processing x requests per second or the number of concurrent jobs the asynchronous Textract API for StartDocumentText is set to in the region.