12 September 2026

11 new posts across 4 engineering blogs — 7 worth your time.

One week of Engineering Reading, 7 stories, as published.

Validating multi-Region DR for Terraform Enterprise with AWS FIS

After an AWS regional service event in us-east-1 made Athenahealth's single-Region Terraform Enterprise deployment inaccessible, the company worked to validate a multi-Region disaster recovery architecture. The post shares their architecture, AWS Fault Injection Service validation approach, and lessons learned.

Why it matters — Using fault injection to validate multi-region disaster recovery architectures helps identify gaps in infrastructure availability before actual regional outages occur.

AWS Architecture · 12 September 2026 · Read the original →

Developer Experience

How we rebuilt Cloudflare Workers’ module registry for Node.js compatibility

Cloudflare has rewritten the module registry in workerd, the core open-source component of the Workers runtime. This rewrite aims to make the registry faster, more standards-compliant, and fully compatible with stable Node.js runtime APIs in a serverless environment.

Why it matters — Rewriting core runtime module registries can achieve full compatibility with established standards like Node.js while improving execution speed.

Cloudflare · 12 September 2026 · Read the original →

Performance

Automatic Key Exchange: faster, post-quantum secure origin handshakes for 45 billion daily connections (and counting)

Cloudflare details how they optimize TLS 1.3 connections to origin servers by guessing the correct key agreement algorithm in the first packet. Making the correct guess avoids a HelloRetryRequest and ensures the handshake completes in a single round trip, even when transitioning to post-quantum secure handshakes.

Why it matters — Committing to the correct key agreement algorithm in the first packet of a TLS 1.3 handshake avoids the latency penalty of a HelloRetryRequest.

Cloudflare · 12 September 2026 · Read the original →

Reliability

Building resilient real-time streaming workers with Amazon DynamoDB leases

This post addresses the challenge of worker failures in real-time streaming services, where a single failure can drop over 100 WebSocket connections and cause minutes of data loss. It details how to build resilient streaming workers that maintain connections using Amazon DynamoDB leases to avoid manual operator restarts.

Why it matters — Using DynamoDB leases to manage worker state prevents minutes of data loss and eliminates the need for manual operator intervention during worker failures.

AWS Architecture · 12 September 2026 · Read the original →

Data & ML

Evolving Pinterest’s Embedding Retrieval Platform

Pinterest discusses the evolution of its embedding retrieval platform to address the 'Billion-Embedding Challenge'. The platform is designed to understand deep semantic signals to power personalized recommendations for home decor and fashion discovery.

Why it matters — Scaling discovery engines to handle billions of embeddings requires evolving retrieval platforms to process deep semantic signals efficiently.

Pinterest · 12 September 2026 · Read the original →

Infrastructure

Building Pinterest’s VLM Serving Stack on NVIDIA Dynamo

Pinterest shares how they built their Vision-Language Model (VLM) serving stack on NVIDIA Dynamo. This infrastructure is critical for supporting the platform's visual search and discovery capabilities by serving complex multimodal models.

Why it matters — Building VLM serving stacks on specialized compilers like NVIDIA Dynamo optimizes the infrastructure required for large-scale visual search.

Pinterest · 12 September 2026 · Read the original →

Infrastructure

1.1.1.1 now supports post-quantum DNSSEC, all 2,420 bytes of it

Cloudflare's 1.1.1.1 resolver now supports and validates DNSSEC signatures made with ML-DSA-44, a post-quantum signature algorithm standardized by NIST. This deployment is part of Cloudflare's roadmap to achieve full post-quantum security by 2029, addressing the challenges of larger signature payloads.

Why it matters — Deploying post-quantum signature algorithms like ML-DSA-44 in DNSSEC is a critical step toward securing internet infrastructure against future quantum threats.

Cloudflare · 12 September 2026 · Read the original →

Preparing for quantum computers while still trying to get multi-region DR to work.

7 stories, every Saturday

Published here every week. Follow by RSS to get it as it lands.

Subscribe by RSS All issues

← Previous issue Next issue →