High Performance Financial Service Integration
summary
Banks must migrate SWIFT MT to MX and adopt ISO-20022 and FpML while handling microsecond trading on FPGA cards, not Java. Real workloads hit ~100000 FIX msgs/s per feed (≈50 MB/s), one client links 28 exchanges producing ~4 TB/hr and ~1.5 GB/s, weeks add up to tens of terabytes. Practical solutions use generated Java bindings, a metadata repo for canonical mapping, EC2 for replay, NoSQL (Mongo/Redis) for persistence, Akka for distribution, and removing XA transactions to regain orders-of-magnitude throughput.
AI-written summary · about
chapters (20)
- 0:00Introduction
- 0:39Agenda
- 8:21Hedge Fund Environment
- 9:13Banking Environment
- 10:25Fixed Messages
- 11:23Trade Capture
- 12:21Middle Office
- 13:24Storage
- 18:18Grid computing
- 19:26FPML
- 21:02Mandatory Rules
- 23:59Swift Message
- 24:28Architecture Diagram
- 27:51Generate Java
- 30:53Add Rules
- 33:24Enterprise Integration
- 40:06Tips
- 44:45Questions
- 47:49How do you measure performance
- 49:41Transaction flow
related talks
52mThe Continued Evolution of High Frequency Trading Systems
45mMigrating Mission Critical Workloads from Mainframe to Open...
38mExpanding Fineract capabilities, a practical example
39mScaling Betfair Exchange to support 160k rps using Apache Kafka
24mBuilding Planet-Scale Payment Platforms
15mPlatform Engineering for FinTech
23mBest of Both Worlds: Building a Mirror-trading Platform on Top of ...
01h 28mHigh Performance Stream Processing