19 system‑design case studies
A widely circulated thread compiled 19 deep system‑design case studies—YouTube, Kafka, Uber ETA, ChatGPT among them—providing practical architecture teardowns for large‑scale systems. The thread is getting traction as a concentrated resource for engineers building high‑scale services. (x.com)
The list is published by Neo Kim under the System Design One banner and appears across both systemdesign.one and a Substack newsletter that bills itself as the System Design Newsletter. (newsletter.systemdesign.one) Entries in the compilation include quantified architecture teardowns such as Tinder’s scale at 1.6 billion swipes per day and Cloudflare’s “55 million requests per second” use case backed by 15 Postgres clusters. (substack.com) Several posts name concrete throughput targets used as framing examples—Uber’s proximity service at about 1 million requests per second and an Uber ETA piece that cites roughly 500,000 requests per second. (substack.com) Infrastructure-focused deep dives catalog platform guarantees and operational metrics, for example Amazon S3’s 99.999999999% durability (11 nines), Zoom’s architecture described around supporting 300 million daily video calls, and Hotstar’s real‑time emoji pipeline delivering billions of events. (substack.com) The publisher’s pieces range from short 5–7 minute Substack explainers to long-form guides—the Live Comment deep dive lists ~9,373 words and a 45‑minute read estimate—while the newsletter archive includes a Jan 28, 2026 explainer on the ChatGPT apps/marketplace (post #116). (systemdesign.one) The thread and its pointers have been scraped and reposted by aggregators (ThreadReaderApp, Rattibha, UnrollNow) and are indexed into the systemdesign.one resource pages, increasing cross‑platform discoverability. (threadreaderapp.com) Substack posts from Apr 26 and May 20 show public subscriber badges of “58,001+” and “63,001+” respectively on the systemdesignone account, indicating measurable audience growth between those posts. (substack.com)
