Low-Latency Trading Technology From FastPanda
All what you need for high-frequency algorithmic trading
Technology
Typical components
of an algorithmic trading platform:


- ticker plant

- algo-hosting
  servers

- risk management
 server


- controlling  
  console(s)

- FIX gateways

Presently there are three major approaches/technological solutions with respect to design of high-frequency algorithmic trading systems: pure software (traditional)  systems, hardware-based systems (financial appliances) and hybrid solutions where system functionality is distributed between software and specialized hardware component(s).

Overall hardware-based and hybrid systems are more efficient compare to traditional pure software-based solutions. However, typically they are less flexible and usually cost of these systems is much higher than traditional software-based solutions. In addition development and debugging of hardware-based systems takes much longer time then for software-based solutions.

On other hand, wide spread popularity of high-performance multi-core computers coupled with deployment of by-pass network adapters or InfiniBand fabric for distributed high-performance computing, using OpenCL to offload heavy calculations to GPUs allows to have a very efficient implementation of software-only trading platform that can easily compete with hardware-based products.