Market data volumes have skyrocketed over the past 10 years (with enormous peaks observed by the Options Price Reporting Authority) reaching over 1.9 million messages per second, challenging all the decoding devices. NASDAQ is also reporting peaks at up to 500,000 messages per second and several million active orders in the market book.
Until five years ago, software technology (particularly Feed Handlers overlaid onto multi-core processors) was used as an efficient way of capturing, decoding and disseminating market data to the client application. Only recently have hardware-accelerated solutions (such as FPGA) emerged from some vendors, allowing them to further reduce decoding times.
Despite hardware-accelerated devices providing a very effective response to ever increasing market data volumes, they have their limits, such as difficult integration with client applications and the lack of instrument management. But combined with powerful software, those limits can be overcome. QuantHouse’s Research & Development team has created QuantBOX, which combines the advantage of hardware capabilities with our best of breed software solution - QuantFEED.
The QuantBOX solution
QuantBOX combines a FPGA-based Market Data Offload Engine (MDOE) with cutting-edge QuandFEED software technology. This mix of technologies provides both outstanding latency and ease of use, in order to access market data as fast as possible.
The Market Data Offload Engine is used to offload the network stack processing (IP/UDP) as well as market data processing from the software (protocol decoding, reordering and enrichment). The hardware accelerator offers the traditional zero copy / kernel bypass functionality to further reduce the latency and avoid unnecessary kernel context switching.QuantBOX has been developed to optimize the critical path while still providing greater system flexibility. Latency is further optimized when the solution is deployed within our data center.
High level of normalization:
Clients benefit from a high level of data normalization including static and dynamic data. High level object oriented API and value add: The normalized data is accessible via one Application Programming Interface (API) and allows easy integration with applications available in Java, C# or C++ through Multicast, TCP/IP or IPC. Furthermore, the API offers publish, snapshot and partial/full subscription functionalities.
High Flexibility and Reliability:
QuantHouse’s market data solution offers huge flexibility. From the API clients can manage their order book, the recovery system and fail over. From an architecture point of view, clients can either have the solution as an appliance or directly collocate their trading strategies on the same server.
Low latency sensitive clients will benefit from a solution which brings market data processing times under the single-digit microsecond and a high throughput with more than 2M messages per second. QuantBOX is a resilient and scalable solution, which is both fully integrated and managed.
- Main Features:
- Data normalized in 2µs
- Full depth order book sorted in single digit µs
- Normalized market data
- Referential data
- Market by order
- Market by level
- Market data snapshot
- Partial subscription
- Advanced API (C#, C++, Java)
- 24/7 managed service li>Hot failover backup