TSTejasvi S Jainintejasvisjain.hashnode.dev·Jun 16 · 12 min readHow the Heck? Do You Count Clicks Without Slowing the Redirect (Kafka)Part 5 — the finale of the "How the heck?" series. Part 4 gave us a replicated Postgres tier behind Pgpool. The last box: click analytics that never slow the redirect down. Part 5 — Async click analy00
TSTejasvi S Jainintejasvisjain.hashnode.dev·Jun 16 · 7 min read How the Heck? Do Read Replicas Actually Work (Postgres + Pgpool)Part 4 of the "How the heck?" series. Part 3 put nginx in front of two stateless app instances. The database is still a single node — both a read bottleneck and a single point of failure. Let's fix th00
TSTejasvi S Jainintejasvisjain.hashnode.dev·Jun 16 · 6 min readHow the Heck? Does One nginx Balance Two App InstancesPart 3 of the "How the heck?" series. By the end of Part 2 we had a working single-node shortener (create + redirect) backed by Postgres. Now we containerize the whole thing and scale the app tier beh00
TSTejasvi S Jainintejasvisjain.hashnode.dev·Jun 16 · 16 min read How the Heck? Do You Go From Empty Folder to Working Shortener (Spring Boot + Postgres)Part 2 of the "How the heck?" series, where I build the URL shortener everyone designs in interviews — by hand, the slow way. In Part 1 I derived the architecture (load balancer → stateless app tier →00
TSTejasvi S Jainintejasvisjain.hashnode.dev·Jun 16 · 6 min readHow the heck ? do you actually build the URL shortener everyone designs?"Design a URL shortener" might be the most common system-design interview question there is. We've all done it: draw a load balancer, a couple of app servers, a primary-replica database, sprinkle in a00
