Transactions Motivation
Concept. A transaction is a multi-step database operation that must run as if it were a single atomic unit, even when thousands of other transactions are running on the same data simultaneously.
Intuition. When 3.5 million Taylor Swift fans hit Ticketmaster at once, each fan's purchase (check inventory, charge card, mark seat sold) must complete entirely or not at all, with no two fans accidentally buying the same seat. The database needs a unit of work that's bigger than a single SQL statement.
Case Study: Taylor Swift Concert Sales on Ticketmaster
The Scale of the Problem
52
Show Dates
3.5M
Fans Pre-registered
2M
Tickets Sold
in a single day
3.5B
System Requests
in a few hours
What Went Wrong?
Concurrent Requests Broke the Site
- Cancelled tickets, reset to later in the day
- 5 hour wait times
- Bots took over the system
- Massive system failures under load
How That Compares
Ticketmaster
3.5B
System requests in a few hours
4x their previous record
Peak transactions/second (TPS): Not reported
VISA Network
110M
Transactions/day average
100K transactions/second at peak
Global payment processing
Amazon Prime Day
105M
DynamoDB requests/second
Sub-10ms response times
288B total transactions/day
The 100× Scale
The whole stack (concurrency, hardware, and the data structures underneath) has 100×'d in each generation. The current era's challenges (and what you'll be solving) sit on top of all three.
| Era | Examples | Concurrent users | Hardware | Data structures |
|---|---|---|---|---|
| Classic (1990s–2000s) | Banks, Walmart, Oracle | ~1,000s/sec | HDD · 1–4 cores · 1 GB RAM | B-Trees · 10K writes/sec |
| Modern (2000s–2020s) | Gmail, Facebook, Roblox | ~10M/sec | SSD · 128+ cores · 1 TB RAM | LSM Trees · 1M+ writes/sec |
| Future (2030s+) | AI agents, serverless apps | Millions of agents/sec | NVMe / CXL · 10K+ AI cores · 1 PB | Learned indexes · ?? |
Where the Field Is Heading
You're entering the field at the perfect time.
Transaction processing is where the innovation and investment is heading. Master this now, lead the future.
What We'll Learn
Concurrency
Millions of users trying to buy the same ticket.
Consistency
Ensuring no double-bookings or lost sales.
Availability
Keeping the system running under extreme load.
Scalability
Handling 100× normal traffic spikes.