Transactions Motivation
Case Study: Taylor Swift Concert Sales on Ticketmaster
The Scale of the Problem
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
Why Study Taylor Swift Concert Sales?
Putting Scale in Perspective
Ticketmaster
System requests in a few hours
4x their previous record
Peak transactions/second (TPS): Not reported
VISA Network
Transactions/day average
100K transactions/second at peak
Global payment processing
Amazon Prime Day
DynamoDB requests/second
Sub-10ms response times
288B total transactions/day
On Complexity
Taylor Swift Tickets
Same seats, millions want them
- Limited inventory (same seats)
- Extreme contention
- Everyone wants exact same thing at same time
- Concentrated time/geography
VISA Network
Global payments, strict consistency
- Financial regulations (zero tolerance)
- Cross-border coordination
- Real-time fraud detection
- 24/7 uptime requirements
Amazon Prime Day
Different products
- Millions of unique items
- Low contention per item
- Distributed load naturally
- Predictable scaling patterns
Why Study This?
More fun than payment processing, and more challenging than e-commerce.
Why Study Transactions
The 100x Scale
Classic Era (1990s-2000s)
Banks, Walmart, Oracle
- Key Problems: Nightly batch processing, Regional scale
- Products: ATM networks, ERP systems
My Era (2000s-2020s)
Gmail, Facebook, Instagram, Roblox
- Key Problems: Real-time updates, Global distribution
- Products: Social networks, Cloud services, Online gaming
Your Era (2030s+)
AI Agents, Serverless, Agent Networks
- Key Problems: Agent coordination, Serverless multi-tenant workloads
- Products: Global agent networks, Extreme multi-tenant apps
Hardware Revolution
2000s Hardware
HDDs, Single cores, MB RAM
Limited what you could build
2020s Hardware
SSDs, Many cores, TB RAM
Enables anything my gen imagined!
2030s Hardware
Cloud-Native, AI Accelerators, CXL Memory
Enables new architectures!
Data Structure Evolution Unlocks Scale
2000s: B-Trees
Oracle, DB2, SQL Server
- Key Problems: Disk-optimized reads, ACID guarantees
- Trade-offs: Slow writes, complex updates
2020s: LSM Trees
Spanner, DynamoDB, RocksDB, Cassandra
- Key Problems: Write amplification, SSD optimization
- Trade-offs: Fast writes, eventual consistency
2030s: Learned Indexes
AI-Native DBs, Vector databases, Learned structures
- Key Problems: Hardware-accelerated data structures, High-dimensional search
- Trade-offs: Training overhead vs index size
You're Building the Future
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.
♂ Extreme Concurrency
Handling massive, unpredictable traffic spikes gracefully
Perfect Consistency
Zero conflicts, zero data loss, zero compromises
Instant Recovery
Self-healing systems that minimize downtime
Global Distribution
Geographically dispersed multi-region systems
What We'll Learn
Real-World Transaction Challenges
1⃣ Concurrency
Millions of users trying to buy the same ticket
2⃣ Consistency
Ensuring no double-bookings or lost sales
3⃣ Availability
Keeping the system running under extreme load
4⃣ Scalability
Handling 100x normal traffic spikes