I have a datadog on site coming up soon and I\'ve seen a few people posted this as their system design question on glassdoor\n\n**Description: Design an app in the style on reddit r/place.**\n\nThere is a NxM pixel grid where users can select a single pixel and place a color\n\nUsers can pick from 15 colors to place on the grid\n\nOnce a user has placed a pixel they can\'t place another one for 5 minutes\n\nPixels can be overwritten by other users\n\n**Inital Requirements**:\n* grid is 100x100\n* We should support at least 100,000 simultaneous users.\n\t\n**Follow up**: make the grid 1000x1000\n\nI\'m able to make this work for 100x100 because the data is relativley small but moving to 1000x1000 I\'m stuggling with how to model the grid with 1 million pixel points \n\n\nThere are some great resources to show how reddit actually handled this\n* [reddit eng blog](https://redditinc.com/blog/how-we-built-rplace)\n* [redis conf](https://www.youtube.com/watch?v=3-Xqghoki70&ab_channel=Redi)\n\nBUT They all seem to rely on [redis bitfield](https://redis.io/docs/latest/commands/bitfield/).  This is a great usage of the redis bitfield but it feels weird bringing up this very specific tool during a system design interview and hinging the whole system on it\n\n

I have a datadog on site coming up soon and I\'ve seen a few people posted this as their system design question on glassdoor\n\nDescription: Design an app in th