Portfolio ↗

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
All AI & LLM Systems8 Data Platforms12 Distributed Systems10 Platform & Infrastructure24 Integration & APIs11 Security & Identity8 Reliability & Operations11 Cost & Efficiency7
Cost & Efficiency 14 Oct 2026

The proxy beside every process: when the sidecar's bill came due

Reconstructs the cost of putting a proxy beside every workload from the teams that measured it: Istio's own versioned benchmarks, the MeshInsight study, Meta's library-first counterexample, and the incident re…

What surprised meThe organisations with the most traffic never paid the tax: Meta routes 99% of RPCs through an in-process library to save hundreds of thousands of ma…
29 sources · 26 organisations · 3 postmortems 29 min read
service-meshsidecaristioenvoykubernetes
Platform & Infrastructure 13 Oct 2026

When one platform runs everything: eleven years of Monzo's architecture

A single-company dig, reconstructed from Monzo's own incident reports and engineering posts, its conference talks, and the archive banners and commit dates in its public repositories. It gives an architect the…

What surprised meThe published justification for a second platform on a second cloud is not that clouds fail: Monzo states that a bug in its own code or processes is …
31 sources · 19 organisations · 3 postmortems 44 min read
monzomicroserviceskubernetesresiliencemulticloud
AI & LLM Systems 13 Oct 2026

When the model provider goes down

Reconstructs the consumer-side model gateway from the builds Uber, Grab, Wealthsimple, Bloomberg, LiteLLM, BricksLLM and OpenRouter published, then reads the layer against the outage record: OpenAI's December …

What surprised meThe layer is sold as outage insurance, yet the only gateway-operator postmortem in the corpus is about the gateway's own caching dependency, and the …
26 sources · 22 organisations · 4 postmortems 31 min read
llm-gatewayfailovermulti-providerrate-limitingincidents
Data Platforms 12 Oct 2026

The data survives the upgrade. The database around it does not.

Reconstructs major-version database upgrades from GitLab's public change and incident tracker, Zalando's Spilo automation, Supabase's platform docs and PostgreSQL's own commit history. A reader leaves with the…

What surprised meThe engine itself committed and then reverted the fixes for two of the four dropped-state traps (sequence replication in 2022, per-index collation tr…
22 sources · 8 organisations · 5 postmortems 31 min read
postgresqlpg-upgradelogical-replicationcutovercollation
Distributed Systems 12 Oct 2026

The object store became the quorum: ten years of Elasticsearch, read from its own repository

A decade of one search engine's coordination design, reconstructed from two complete git histories, the design documents committed inside them, and a known-issues file that doubles as an incident archive. It s…

What surprised meElasticsearch shipped master election through an object-store register in April 2023 on a compare-and-exchange protocol built out of multipart upload…
26 sources · 2 organisations · 3 postmortems 36 min read
object-storageconsensuselasticsearchopensearchfencing
Reliability & Operations 11 Oct 2026

Evacuating the region: who can actually leave, and what it costs

Reconstructs regional evacuation from GitLab's public DR blueprints, runbooks and gameday timing ledger, Meta's Maelstrom and Taiji papers, Netflix's active-active record, Datadog's failover-safety work, and t…

What surprised meGitLab's regional-recovery change template pre-prints a 96-hour downtime component, and the biggest failover disaster in the record (GitHub 2018) cam…
26 sources · 15 organisations · 6 postmortems 35 min read
disaster-recoveryfailovermulti-regiongamedaysrto-rpo
Integration & APIs 11 Oct 2026

What the vendor cannot host, you will maintain: ten years of Alibaba's published middleware

A decade of Alibaba's published middleware, for architects choosing which parts of their own system to delegate to a large company's components. The record shows the servers (registry, broker, coordinator) sti…

What surprised meThe final act of a 15-year JSON library was a jar uploaded to Maven Central at 07:24 on 29 July 2026, hours before its repository went read-only, wit…
36 sources · 9 organisations · 6 postmortems 33 min read
open-sourcedependenciesservice-governancesupply-chaindeprecation
Security & Identity 10 Oct 2026

Rotate the key while nothing is wrong

Reconstructs trust-anchor rotation from the systems that do it on schedule (the DNS root, Entra ID, Sigstore, OpenSSH, Let's Encrypt) and the incidents where it was absent or late: Storm-0558's seven-year-old …

What surprised meMicrosoft stopped rotating MSA signing keys in 2021 because a manual rotation had caused a major outage; the fix for an availability incident left th…
34 sources · 30 organisations · 6 postmortems 34 min read
key-rotationtrust-anchorpkidnssecsigning-keys
Integration & APIs 10 Oct 2026

Swapping the system behind the name: ten years of LinkedIn

LinkedIn wrote Kafka, published it, and in 2025 announced its replacement. Over the same decade it also left its own RPC framework, its own service discovery format, its own key-value store and its single stre…

What surprised meXinfra, the layer that lets topics move off Kafka, was first built as znodes inside LinkedIn's Kafka fork in October 2023, twenty months before North…
28 sources · 15 organisations · 3 postmortems 33 min read
kafkamigrationcompatibilityrpcservice-discovery