Engineering & Guides

System Design

21 articles on system design.

Internal developer portal dashboard on top of an internal developer platform engine

Internal Developer Portal vs Internal Developer Platform: What Engineers Need to Know

Internal developer portal vs internal developer platform: the portal is the interface developers click, the platform is the engine that does the work. Here is the difference and which to build first.

8 min read Jul 4, 2026Read

Editorial illustration of a single large monumental monolith block standing confidently in the center, glowing with warm light to represent one unified application, with streams of glowing data flowing into it from many directions to suggest enormous traffic at scale, and a few small supporting glowing server racks, a cache layer, and a queue around its base quietly absorbing the load, in a calm teal and deep blue palette with warm amber accents.

System Design

Scaling a Monolithic Architecture to Billions: PostgreSQL, Django, No Kubernetes

How some of the largest products ever built scaled a monolithic architecture to enormous size with PostgreSQL, Django, Memcached, and a task queue, no Kubernetes or microservices required.

10 min read Jun 30, 2026Read

Editorial illustration of three network routing components that all sit in front of servers: a single glowing gateway front door where data arrives, a splitter that fans one stream of traffic into parallel lanes across identical server racks like a load balancer, and a smart junction routing labeled pipes to differently shaped service blocks like an api gateway, connected left to right by flowing arrows of data in a teal and deep blue palette with amber accents.

System Design

API Gateway vs Load Balancer vs Reverse Proxy

They all route traffic, but a reverse proxy, a load balancer, and an API gateway solve different problems. Here is how to tell them apart and when to use each.

7 min read Jun 29, 2026Read

Editorial illustration of choosing between Redis and Memcached. On the left a richly structured glowing data engine of stacked layers and geometric shapes with a small disk underneath represents Redis with its data structures and persistence. On the right a simple repeated grid of identical fast cache boxes connected by parallel lines represents Memcached as a plain horizontally sharded blob cache. A decision fork in the center suggests the tradeoff between power and simplicity.

System Design

Redis vs Memcached: When "Just a Cache" Is the Right Answer

Redis vs Memcached, decided by workload: how data structures, persistence, and clustering set them apart, and when a plain cache is the right answer in a system design interview.

10 min read Jun 28, 2026Read

Editorial illustration of latency versus throughput: a single stopwatch racing down a thin fast lane represents latency, the time for one trip, while a wide multi lane highway full of parallel data streaks represents throughput, many operations per second, with a ladder of hardware tiers from cpu and memory to disk and a networked globe between them

System Design

Latency vs Throughput: Numbers Engineers Must Know

Latency vs throughput, made clear: the latency numbers every engineer should know, how bandwidth differs, and how to estimate storage and QPS in a system design interview.

8 min read Jun 27, 2026Read

Editorial illustration of back of the envelope capacity estimation: an engineer at a whiteboard sketches quick rough numbers and arrows flowing from user icons to a speed gauge, storage disks, a cache chip, and streaming network lines, conveying fast approximate math for a system design interview

System Design

Back of the Envelope Calculation: 5 System Design Examples

A back of the envelope calculation, made simple: five worked examples turning daily active users into QPS, storage, cache, and bandwidth for a system design interview.

6 min read Jun 26, 2026Read

Bloom filter concept illustration showing a bit array with hash function arrows pointing to specific positions

System Design

Bloom Filters Explained: When "Probably Yes" Is Good Enough and Saves 99% of Memory

Bloom filters trade a tiny false positive rate for 95% memory savings. Here is how they work and where they show up in Cassandra, Redis, and Chrome.

12 min read Jun 2, 2026Read

Consistent hashing ring diagram showing virtual nodes distributed across physical servers

System Design

Consistent Hashing Explained: Virtual Nodes and Why They Matter for Distributed Systems

Learn how consistent hashing with virtual nodes solves load balancing and data distribution in distributed systems like Cassandra, DynamoDB, and Redis Cluster.

12 min read Jun 1, 2026Read

Internal developer platform architecture illustration

System Design

How to Build an Internal Developer Platform: Architecture, Tools, and Lessons Learned

A practical guide to building an internal developer platform — the five-layer architecture, tool selection between Backstage, Port, and Humanitec, phased rollout strategy, and hard lessons from production IDPs.

12 min read Jun 1, 2026Read