Software Architecture and System Design News
Latest curated articles from top engineering blogs
Shopify
AWS1001 articles
Evolving System Design Paradigms for 2026: Beyond Traditional Architectures
This article discusses the shift in system design paradigms, moving beyond traditional architectures of Load Balancers, API Gateways, and microservices to address the challenges of 2026. It emphasizes a return to first principles, incorporating AI-aware design, and focusing on operational maturity, cost reasoning, and blast radius reduction. The core message is about designing efficient, sustainable systems that handle failures gracefully and integrate emerging technologies like AI effectively.
Cloudflare's Cache Transcoding for Storage Optimization
Cloudflare is prototyping Cache Transcoding, a system that uses Zstandard compression to reduce storage requirements for eligible cache content (HTML, JSON, CSS, JavaScript). This approach aims to provide petabytes of additional effective cache capacity by compressing content once upon entry and decompressing when served, demonstrating a trade-off between CPU cycles and storage/bandwidth savings.
Understanding and Mitigating Cache System Failures
This article explores common failure modes in distributed cache systems, including the thunder herd problem, cache penetration, cache breakdown, and cache crashes. It provides practical solutions and architectural considerations to enhance cache resilience and maintain database stability in high-traffic environments.
Designing a Real-Time Financial Risk Management System with Automated Liquidation
This article delves into the architectural considerations for building a real-time financial trading platform, specifically focusing on its risk management engine. It explains the mechanics of maintenance margin, short-selling, and automated liquidation logic, highlighting how a system like VTrade handles real-world financial boundaries and ensures the platform's stability. The discussion emphasizes the need for continuous, high-frequency risk evaluation and programmatic control over leveraged positions to prevent cascading losses.
AWS Architecture Blog·15h agoBuilding Resilient Real-Time Streaming Workers with DynamoDB Leases
This article details an architecture for managing long-lived WebSocket connections in a fleet of real-time streaming workers, addressing challenges like worker failures, deployments, and scaling. It focuses on using Amazon DynamoDB's conditional writes to implement a distributed lease pattern for connection ownership, ensuring high availability and automatic failover without external coordination services. The pattern significantly reduces recovery time and manual intervention.
AWS Lambda SnapStart for Container Images: Optimizing Cold Starts for Heavy Dependencies
AWS Lambda SnapStart's extension to container images significantly reduces cold start times for functions with large dependencies, eliminating a prior trade-off. This feature enables developers to package functions exceeding 250 MB using container images (up to 10 GB) without incurring multi-second startup delays, previously a major challenge for applications relying on libraries like pandas and numpy.
Introduction to Software Design and Architecture Principles
This article provides an essential overview of software design and architecture, emphasizing its critical role in successful software projects. It outlines core principles, common architectural patterns, and the importance of considering non-functional requirements to build robust and scalable systems.
Designing Uber: Geospatial Matching, Live Location, and Surge Pricing
This article outlines the core system design challenges behind Uber, focusing on how to efficiently match riders with nearby drivers using geospatial indexing, manage real-time location updates at scale, and implement dynamic surge pricing. It delves into the architectural considerations for handling high request volumes and low-latency data processing essential for a ride-sharing platform.
Migrating Microservice Communication from JSON to Protobuf for Performance and Efficiency
This article outlines a strategy for migrating internal microservice communication from JSON to Protobuf to address performance bottlenecks at scale. It details the reasons why JSON degrades under high load, highlights Protobuf's efficiency benefits, and provides a zero-downtime migration blueprint. The core focus is on improving CPU utilization, reducing network payload size, and leveraging gRPC over HTTP/2 for internal service meshes.
Pinterest's Evolution of Embedding Retrieval for Billions of Items
This article details how Pinterest scaled its in-house distributed search platform, Manas, to handle tens of billions of embeddings efficiently. It focuses on architectural changes to optimize cost, scalability, and flexibility, particularly through quantization, SSD-based serving for ANN queries, and the adoption of multi-embedding retrieval models for richer semantic search.
Uber's ETA System: Predictive Traffic Forecasting with DeepETA
This article dissects Uber's sophisticated Estimated Time of Arrival (ETA) system, detailing its architecture from raw GPS data collection to a deep learning model for predictive traffic forecasting. It highlights how Uber leverages its vast fleet as a real-time sensor network, segments roads, and employs an AI model, DeepETA, to predict future traffic conditions for accurate ETAs, significantly impacting user experience and business metrics.
NVIDIA Personal AI Router for Local Distributed AI Inference
The NVIDIA Personal AI Router (PAIR) enables distribution of AI inference tasks across multiple local machines, addressing bottlenecks in multi-agent AI workloads. It acts as a proxy, routing requests to available nodes running local inference services like Ollama or LM Studio, thereby maximizing local compute capacity without merging GPUs or pooling VRAM. This system design allows for more efficient local AI processing by parallelizing independent inference calls.