Slack: Channel-Based Messaging & The Flannel Edge Cache

Slack powers real-time workplace collaboration for tens of millions of daily active enterprise workers. Unlike consumer chat apps (where users talk in small 1-to-1 threads), Slack operates around Channel-Based Workspaces:

• An enterprise workspace can contain over 50,000 employees.
• A single user routinely belongs to 50 to 100 public and private channels.

The "Client Boot Storm" Challenge:

When an employee opens the Slack desktop app in the morning, the client cannot simply display a blank screen. It must instantly render:

• Metadata for all 50 channels (names, topics, pins, permissions).
• Workspace membership roster (10,000 user profiles, display names, avatars).
• Real-time online/away presence status for every colleague.
• Unread message counters and mention badges for every channel.
• Custom workspace emojis and integration tokens.

Executing hundreds of SQL queries against central MySQL databases for every client launch created massive Client Boot Storms at 9:00 AM across global time zones, overwhelming database connection pools and causing minutes of startup lag.