Energy & commodities trading
Ultra-low-latency tick capture with the Disruptor pattern
4storage targets fed from one capture path
Real-time capture of orders and trades, delivered to four storage targets without slowing the hot path.
Context
Alongside the historical collection, the desk needed live tick data, both orders and trades, from Trayport for real-time strategies, analytics and monitoring.
The challenge
- Keep latency low and stable during market bursts.
- Deliver every tick to Kx Insights, TimescaleDB, QuestDB and Azure Blob Storage.
- Stop one slow consumer from holding back the others.
What we did
- Designed the capture pipeline around the Disruptor pattern: a pre-allocated ring buffer with lock-free hand-offs between stages.
- Gave each storage target its own consumer, so a slow sink applies back-pressure locally instead of to the feed.
- Delivered the services as containers through Podman and Harbor.
Outcome
- A real-time tick feed available to algo traders alongside the historical data.
- An architecture where adding a storage target does not touch the hot path.
Technologies
Next case study
Audit and fills data for trading surveillance
Audit and fills data for trading surveillance