Digs

Searching X architecture returns material written to be found by that query, which means tutorials and vendor pages. The material worth having was written by an engineer explaining a real system to their peers, and it uses different words. A dig is what comes back from going after that layer on one topic: the architecture reconstructed across real systems, the decisions and the conditions that flip them, what actually broke, and numbers you can plan against. Every claim is graded, linked and dated.

91 field guides 2462 graded sources 422 postmortems 8 categories
Platform & Infrastructure 07 Sep 2026

The half-life of in-house infrastructure: a decade of Uber, read from its own repositories

A decade of one company's platform decisions read from the artefacts nobody edits on purpose: archive banners, last-commit dates, closed-unmerged pull requests and operator-filed incident reports. It gives an …

What surprised meThe Uber-originated software still shipping in 2026 is almost entirely the software Uber gave away, while several systems it kept are dead without sa…

26 sources · 8 organisations · 4 postmortems 33 min read
open-sourcebuild-vs-buydeprecationgovernanceuber
Integration & APIs 07 Sep 2026

The retry you are counting on was never promised

Reconstructs the sender's side of webhook delivery from GitLab's incident reviews and shipped source, GitHub's staff answers, three open-source delivery services and the Standard Webhooks specification. An arc…

What surprised meGitHub and GitLab, whose webhooks trigger most of the world's CI, do not retry failed deliveries at all; the at-least-once guarantee everyone assumes…

24 sources · 10 organisations · 4 postmortems 28 min read
webhookseventingretriescircuit-breakerdelivery
Data Platforms 06 Sep 2026

A million clients, four hundred connections

Reconstructs the two-line connection-rationing architecture that GitLab, Notion, Figma, OpenAI, Cloudflare and Supabase all converge on, with the measured cost of idle connections, the ceilings of a single-thr…

What surprised meThe mature deployments no longer engineer connection reuse at the pool; they engineer admission control there, because the pool is the one chokepoint…

33 sources · 26 organisations · 5 postmortems 30 min read
postgresconnection-poolingpgbouncercapacityadmission-control
Platform & Infrastructure 06 Sep 2026

The monolith that could not be split: ten years of GitLab architecture, 2016-2026

A decade of one company's architecture, reconstructed entirely from primary artefacts: design documents with status fields, twenty-eight numbered decision records, a rejected/ directory, closed merge requests …

What surprised meThe flagship horizontal-scaling programme, four years and twenty-eight decision records in, was put on hold in its entirety and replaced by one whose…

32 sources · 2 organisations · 7 postmortems 32 min read
monolithshardingpostgresmulti-tenancygitlab
Distributed Systems 05 Sep 2026

Time cannot go backwards, and other lies production believed

Two postmortems (Cloudflare 2017, Azure leap day 2012), three Linux kernel commits, the Go and Rust standard-library design records, Kubernetes leader election, and the clock architectures of Google, Meta, AWS…

What surprised meThe industry's fix for the leap second is to make every clock deliberately wrong, and no two operators make them wrong the same way: Google smears 24…

23 sources · 14 organisations · 2 postmortems 27 min read
clocksmonotonic-timeleap-secondfencing-tokenshybrid-logical-clocks
Platform & Infrastructure 04 Sep 2026

When name resolution fails, the record was wrong, not the server

Six published incidents (AWS October 2025, Meta 2021, Slack 2021, Cloudflare 2025, Salesforce 2021, Roblox 2021) plus the build records at Stripe, Kubernetes, musl and Netflix, reconstructed into a reference r…

What surprised meIn five of six major naming outages the query-serving infrastructure stayed healthy; what failed was the automation and change process that decides w…

28 sources · 26 organisations · 6 postmortems 32 min read
dnsservice-discoverycontrol-planecachingttl
Data Platforms 03 Sep 2026

The disks never disappear: what "diskless" systems actually delete

Reconstructs the zero-disk pattern across nine production systems (WarpStream, Kafka's KIP-1150, Confluent Freight, Neon, turbopuffer, Datadog Husky, SlateDB, AutoMQ, Snowflake's ancestor) from their own desig…

What surprised meEvery 'diskless' system in the public record kept a small, strongly consistent, often disk-backed ordering core; Aiven's diskless Kafka removed the b…

24 sources · 20 organisations · 4 postmortems 30 min read
object-storages3kafkadurabilitycost
Distributed Systems 02 Sep 2026

A cache miss is a protocol, not an event

Reconstructs the miss protocol behind every production cache from Facebook, Slack and Wikimedia postmortems and the implementations at Google, Instagram, Discord, DoorDash, Netflix, memcached, Rails, nginx, Va…

What surprised meIn the worst published cache outages the stampede was triggered by the machinery that manages the cache, not by an expiring key: Facebook's repair pa…

26 sources · 22 organisations · 3 postmortems 26 min read
cachingcache-stampedethundering-herdmemcachedstale-while-revalidate
Integration & APIs 01 Sep 2026

A timeout is not an answer: making retried requests safe

A client that times out on a mutating call holds no evidence about whether it happened, so it must resend, and something has to make the resend harmless. This guide reconstructs that machinery from Stripe's pu…

What surprised meThe machinery that guarantees at-most-once fails silently like any other software: Kafka 3.0.0 and 3.1.0 shipped idempotence 'on by default' while a …

28 sources · 25 organisations · 4 postmortems 33 min read
idempotencyretriespaymentsapi-designexactly-once