Backend & System Design Preparation

A rare edge case has allowed customers to check out products at an outdated, incorrect lower price immediately after a global price update was deployed.

A critical severity bug was just reported: some users are seeing private account information belonging to completely different users when they load their authenticated dashboard.

A sophisticated attacker is successfully taking over user accounts despite users having strong passwords. The authentication service hasn't seen brute force attempts, yet accounts are being compromised continuously across different IP regions.

It is Diwali Night, the biggest event of the year. At 7:00 PM, traffic spiked to 3.5x normal levels as thousands of users tried to place dinner orders simultaneously. The database CPU hits 100% and connection pools are exhausted, causing a catastrophic cascading failure. A legacy monolithic Node.js API connects directly to a single Postgres Primary instance.

A major celebrity joined the platform and accumulated millions of followers in hours. However, their follower count on the profile page often shows wildly fluctuating numbers, occasionally dipping lower before jumping back up.

After a 30-minute planned maintenance window, the application comes back online and instantly goes down again. The authentication service is completely overwhelmed despite being previously over-provisioned.

A weekly payout cron job ran last night, but this morning 10,000 users never received their funds. All internal monitoring dashboards are completely green, showing zero errors or exceptions.

A scheduled background cron job responsible for capturing monthly subscriptions crashed mid-execution. When operations manually restarted the job today, thousands of users were charged a second time.

Over the past month, the home feed has been degrading. Today, the P99 latency for the main feed endpoint hit a staggering 10 seconds. Users are staring at infinite loading spinners, and engagement metrics are plummeting. The database read replica IOPS are constantly maxed out. Users expect their home feed to load in under 500ms.

A new viral challenge is sweeping the platform. Millions of users are simultaneously trying to upload their massive 4K video clips. The background processing queue has backed up by 12 hours, and the main upload API is completely unresponsive, throwing 503 errors to all clients.

A single heavy user is causing the entire application to crash periodically, resulting in widespread 503 timeouts for everyone else. Monitoring shows severe memory spikes and CPU saturation closely tied to API requests from this specific user account.

During a massive midnight flash sale, customers rushed to checkout. Afterwards, customer support tickets explode claiming they were billed twice for the same order.