Skip to content

Bedrock Throttling Exception for new user

-1

Hi,

I have been trying to use the model playground and everytime I use bedrock model playground, I get the throttling exception. Note that I have limited the output token to 400 only but this is still happening for any model though I primarily tested for Deepseek. The region I am using is ap-south-1 and though I claim to be a new user, actually it is not since the account was created couple of months back at least. I raised support ticket with AWS to increase quota and I was told that's not possible because I haven't used my existing quota and that makes sense because I have been unable to call any LLM in bedrock. Any help is appreciated.

Regards

asked 7 days ago44 views
3 Answers
1

Seems you hit the account‑level quota initialization rather.

https://docs.aws.amazon.com/bedrock/latest/userguide/quotas-increase.html

EXPERT
answered 7 days ago
EXPERT
reviewed 7 days ago
  • Well that's what I thought. But when I raised support ticket for increasing quota I was told that I have barely used the existing so increase is not possible

1

I understand your frustration with the throttling exceptions in Amazon Bedrock. Here are some key points to help resolve this issue:

Understanding Throttling vs. Quota: Throttling exceptions are different from quota limits. Throttling occurs when you exceed the rate of requests allowed per minute (transactions per minute or TPM), while quotas refer to the total capacity allocated to your account. Even with a low output token limit of 400, you can still hit throttling limits if you're making requests too rapidly.

Troubleshooting Steps:

  1. Check Your Request Rate: Throttling exceptions typically occur when you exceed the allowed requests per minute for your model. Try spacing out your requests with delays between API calls, even in the playground.

  2. Verify Model Access: Ensure you have explicitly requested and been granted access to the specific models you're trying to use (like DeepSeek). Model access must be enabled before you can use them, even in the playground.

  3. Review Service Quotas: In the AWS Service Quotas console, check your current limits for Amazon Bedrock in ap-south-1. Look at both the "Requests per minute" quotas and any model-specific quotas.

  4. Implement Retry Logic: If you're using the API programmatically, implement exponential backoff retry logic to handle throttling exceptions gracefully.

  5. Consider Different Regions: If ap-south-1 has lower default limits, you might try testing in other regions where you might have different quota allocations.

For Quota Increases: While AWS support mentioned you haven't used your existing quota, the throttling issue suggests you might be hitting rate limits rather than capacity limits. You may need to specifically request an increase in "requests per minute" rather than overall capacity quotas.

If the issue persists, provide AWS support with specific details about the exact error messages, timestamps, and the frequency of your requests to help them better diagnose the throttling issue.

answered 7 days ago
EXPERT
reviewed 7 days ago
0

I am hitting the service once a week almost. How come I am breaching request per minute?

answered 5 days 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.