Textract faster on some days rather than others

Question

Previously about 2 weeks ago we were experiencing around 10-15 seconds per pdf when querying textract to translate our pdf. All the sudden now with the same code it takes give or take 5 minutes. Some key things I would like to point out are the PDFs are no larger and the code has not changed since then. The language I'm using is python and the library is boto3. Currently we are using the start_document_text_dection function that points to a pdf located in our s3 bucket. Once we determine the PDF is done reading (JobStatus changes from IN_PROGRESS), we then use the get_document_text_dection function to wrap the process up.

Accepted Answer

Recommend that you create a service ticket with support to dig deeper into your issue and the region you are operating within. They can then examine if there was a regional latency issue and can provide this information back to you. You can also try leveraging an alternative region to check and see if your latency returns to 10-15 seconds but beware that not all regions charge the same amount per API.

Textract faster on some days rather than others

Relevant content