Member-only story
This blog is just a rant, born out of frustration after discovering an unexpectedly high bill for AWS Bedrock Knowledge Base (KB) and OpenSearch Serverless. While serverless architectures promise scalability and cost-efficiency, I’ve learned the hard way that ‘serverless’ and ‘on-demand’ are marketing terms that don’t necessarily translate to true pay-per-use pricing.
After testing the service briefly and developing minimal features, I was shocked to discover a nearly $200 bill for OpenSearch Serverless. Here’s what happened and why AWS Bedrock’s serverless architecture doesn’t deliver on the pay-per-use ideal.
I know I’ve made some mistakes along the way by not being thorough. But in my defense, these were mistakes anyone could have made, such as:
From a frustrated user on Reddit:
I have the same issue — used Bedrock to set up a KB for RAG which embeds less than 2Gb worth of PDFs and ran ~10 queries to the LLM. The LLM (Claude) and S3 are dirt cheap, but I’ve already racked up 10$ just from setting this up. It’s supposedly serverless, but already racked up 21 hours for indexing and 21 hours for search
The Setup: My Experience with AWS Bedrock KB
As someone who had been exploring AWS’s RAG capabilities for a while, I was hesitant to try OpenSearch Service due to its provisioning requirements for EC2 and EBS storage. Even though the service includes a…
