TransUnion: Scaling Data Heavy Transactions w/ Aerospike + Adaptive Map
summary
TransUnion moves billions of identity records on an all‑flash Aerospike cluster and updates nearly 4 billion records daily by using Adaptive Map to split records beyond the 1 MB block limit, pushing business logic into Aerospike Expressions and combining optimistic batch + async writes. Historical loads reach ~60k+ transactions/sec and overall writes exceed 100 billion, batch ops show 99% <90 ms and 99.9% <150 ms, and customer API reads respond in a few milliseconds by reading root indexes then fetching child shards as needed.
AI-written summary · about
chapters (11)
- 0:00TransUnion fraud prevention intro
- 4:31Kafka to Aerospike migration
- 7:07Generation version approach limits
- 8:15Adaptive Map for scalable records
- 11:09Aerospike Expressions optimization
- 12:35Batch operations for performance
- 15:47Adaptive Map performance stats
- 19:45Monitoring batch operations
- 21:17Data model improvements
- 22:44Cloud scaling impact
- 23:08Q&A
related talks
49mJourney with Aerospike: Architecture Best Practices
33mInnovation with Aerospike: How AppLovin Handles Billions of Requests per Day
36maccelerating the digital agenda
35mReplacing Cassandra: A Digital Transformation for the World’s Largest Digital Identity Network
31mUser Attribution at Scale: All-Flash, ½ Trillion Objects
33mUsing Aerospike as a Real-time System-of-Record for Adobe Experience Cloud Services on Azure
22mThe Art of the Possible
29mWhat breaks when Aerospike hits 6 million QPS