How are they connect to S3? are they using a VPC endpoint / NAT? If they are using a VPC endpoint, My recommendation will be the open a support ticket, it's possible that support will be able to look at the network logs.
Another option for the customer is to use pipe input, pipe mode is recommended for large datasets, and it'll shorter their startup time because the data is being streamed instead of being downloaded to your training instances.
Do I have to redownload dataset to training job every time I run a Sagemaker Estimator training job?asked a year ago
How to send own failure info in case of failed SageMaker Training Job?asked 7 months ago
Sagemaker taking an unexpectedly long time to download training dataAccepted Answerasked 4 years ago
Amazon SageMaker - Training Job / Data Wranglerasked 2 months ago
Comprehend Custom classification training taking a lot of timeasked 6 months ago
Can I limit the type of instances that data scientists can launch for training jobs in SageMaker?Accepted Answerasked 2 years ago
Code running slow on Sagemaker notebook instance for the first time it runsasked 2 years ago
Which Amazon SageMaker algorithms can only use GPU for training?Accepted AnswerMODERATORasked 2 years ago
How to checkpoint SageMaker model artifact during a training job?Accepted AnswerEXPERTasked 3 years ago
Sagemaker training instanceasked 4 months ago