# Angular SSR for 40 Million Sessions a Day | Julian Jandl @ Angular Vienna

*Talk | LEAN-Coders, Vienna | 2026-10-01T00:00:00.000Z*

## Description

Most SSR tutorials end at ng add @angular/ssr and deploy. None tell you what happens when you retrofit server-side rendering onto a mature Angular app composing a multi-tenant suite of products, serving real-money betting on data that moves second by second. This talk is the playbook that had to be written because none existed.

The architecture is Live-Data ISR: pre-rendered HTML kept continuously warm against live betting data, served from a cache, with one load-bearing guarantee, cache miss equals today. A page that isn't rendered falls back to the existing CSR experience, never a worse one. That single property is why SSR shipped across 40 brands with zero player-facing incidents, and moved p75 LCP from ~10.5s to ~2.7s on the devices that needed it most.

Then the frontiers you only hit in production: a browser-built codebase reaching for window and localStorage across hundreds of packages; memory leaks a short-lived tab forgives but a long-lived Node process accumulates; the render that did 2 seconds of work then sat idle for 18; ~30 distinct hydration failure patterns distilled into a four-layer diagnostic model; and a 100-pod render farm bottlenecked on TLS and connection handling, not application logic.

The throughline is AI as the force multiplier that made this tractable at all. A custom pattern detector feeds an AI-assisted, package-by-package migration: the agent finds violations, plans the change, applies it as a mechanical edit, runs the tests, and tickets the ambiguous cases for a human. The same rule set powers a merge-request review bot that flags SSR violations before they land, with the canonical fix attached, so ~2,500 contributors who never think about SSR still get caught at authoring time. Both run off a curated, machine-readable knowledge base of what SSR-safe means in this codebase, the context layer that lets an agent give answers specific to the platform rather than generic Angular advice.

You'll leave with a model for fail-safe SSR, a vocabulary for the failure modes you'll hit at scale, and a concrete picture of how AI workflows keep a giant codebase SSR-healthy under a weekly release cadence.

## Slides

> The slides are founde [here](https://push-based.github.io/talk-ng-vienna-011026-ssr-on-scale/1 "Link to the slides") !