Hi Community,

I wanted to open a discussion regarding the current state of running Burr
on AWS and propose an architectural roadmap to improve enterprise readiness.

While the local development experience is excellent, we encounter
significant friction when deploying to production environments on AWS.
Specifically, the polling-based telemetry introduces an observability
latency of 30–60 seconds, which disrupts the debugging loop,particularly
for human-in-the-loop workflows.

Additionally, I’ve observed fragility in our persistence layer
(specifically *UnsupportedOperation*: not seekable errors with S3 streams)
and identified a need to reduce the boilerplate required for Amazon Bedrock
integrations.

*Proposal: Cloud-Native Architecture (BIP-42)*

To address these gaps, I have drafted a comprehensive design proposal to
transition Burr toward an *Event-Driven Architecture*.

*Key Highlights:*

   -

   *Event-Driven Telemetry:* Replace S3 polling with an SQS + EventBridge
   pipeline. This aims to reduce UI latency from ~60s to *<200ms*.
   -

   *Reliable Persistence:* Implement a SeekableStreamAdapter that
   intelligently buffers S3 streams (using RAM with a disk spillover),
   resolving the serialization failures.
   -

   *Native Bedrock Integration:* Introduce a standard BedrockAction class
   to abstract implementation details (streaming, guardrails, throttling) and
   standardize usage patterns.
   -

   *Infrastructure as Code:* A canonical Terraform module to provision this
   securely (VPC-native, private subnets), directly addressing *GitHub
   Issue #391*.

*Review & Feedback*

I have detailed the full technical specification in the document below and
cross-posted it to GitHub Issue #391.


   - * Proposal Document:*
   
https://docs.google.com/document/d/1mZeJz1ERBpBlhQ8uXtsmoiuuiwlrVdUETBdFpFh1IoQ/edit?tab=t.0
   -  *GitHub Discussion:* https://github.com/apache/burr/issues/391

I am particularly interested in feedback from those running Burr in strict
security environments.does the proposed Terraform module structure align
with your deployment constraints?

Regards,

Vaquar Khan

Senior AWS Architect
*Linkedin *-https://www.linkedin.com/in/vaquar-khan-b695577/
*Book *-
https://us.amazon.com/stores/Vaquar-Khan/author/B0DMJCG9W6?ref=ap_rdr&shoppingPortalEnabled=true
*GitBook*-https://vaquarkhan.github.io/microservices-recipes-a-free-gitbook/
*Stack *-https://stackoverflow.com/users/4812170/vaquar-khan
*github*-https://github.com/vaquarkhan

Reply via email to