Newsletter platform inspired by Substack and Beehiiv: readers subscribe, issue archive pages are served from the edge, and sending an issue fans out through a queue to every subscriber.
See why sending to every subscriber from the web server does not scale.
What was missing: Sending an issue should not run inside the web server.
Why that's risky: Sending a million emails ties up the same process that serves readers, so the site slows down on every send.
What gets added: One server that saves signups, shows archive pages and sends each issue to every subscriber itself.
Trade-offs: Simple, but sending and serving fight each other.
Split the paths and serve archive pages from memory.
What was missing: Reads and signups share one path.
Why that's risky: A new issue can take a moment to appear in the cache.
What gets added: An archive service reading from a cache, a subscribe service that stores each signup under a unique email per newsletter, and one database.
Trade-offs: Reads are fast but slightly stale; signups are idempotent so a double click does nothing.
Serve archive pages from the edge.
What was missing: Every archive read reaches our servers.
Why that's risky: A new issue needs its edge copy purged or readers see the old archive page.
What gets added: A CDN in front of the archive.
Trade-offs: A short TTL or purge on publish keeps the archive fresh.
Send each issue to subscribers in batches through a queue.
What was missing: Sending blocks the web servers and bots can fill the list with fake addresses.
Why that's risky: If workers fall behind, a send that should take an hour takes much longer.
What gets added: A send queue that carries confirmation emails for new signups and one batch of subscribers per new issue, workers that send each message and record delivery, a signup rate limiter and monitoring.
Trade-offs: Email providers throttle senders and judge reputation, so workers send at a steady rate and stop sending to addresses that bounce; sending faster risks landing in spam.
You're in the middle of an interview session. Leaving now will end your current attempt.
Explore concept overviews, real-system examples, key tradeoffs, and interview talking points for each roadmap section.
You've conquered this phase. These are the skills you now own: