Skip to main content

Latency

How to reduce latency

There are two main options which can reduce latency:

  • Optimizing connection

  • Optimizing data

Optimizing connection

Colocation

You may choose your latency based on a simple approach: if the connection is closer to dxFeed infrastructure, the latency is less:

Colocation -> Amazon -> Internet

dxFeed offers a colocation option: you can be connected to the nearest data center in order to reduce ping. Check our connectivity page to find the nearest center to you.

Increasing speed

dxFeed can provide a huge amount of data per second. But it is impossible to provide more data than a customer can receive. Thus, dxFeed advises increasing the speed of connection. An acceptable connection speed starts from 1MBit/Sec.

Cross connection

In order to reduce data package losses and disconnects, dxFeed offers a cross-connection. Usually, cross-connection consists of a public and a private connection.

Optimizing Data

For users that are far from our servers and claim to have acceptable bandwidth (not less than 1 MBit/Sec):

  • Breaking down a single request for 1 day of data to multiple requests (for example, to 1 hour of data) and querying them one by one.

  • Breaking down a single request for 1 day of data to multiple requests (for example, to 1 hour of data) and querying several of them in parallel (e.g., 3 parallel flows).

  • Place new frontend servers closer, in order to reduce the chance of packet drops.

Data granularity

Granularity means the level of detail of your data within the data structure. Learn more about granularity.

Granularity usage:

  • Split one day into 24 segments (by hour)

  • Query most recent segments with the highest granularity

  • Query the oldest segments with the smallest granularity

  • Analyze preliminary user behavior