Files
spark/src/sdk

BTS Spark

Our offering consists of a client UI (Spark Client), a high performance professional futures front end, connected via our proprietary protocol over TCP to our Spark Server, which in turn connects to execution venues via their native protocols (typically FIX/TCP for execution and some proprietary protocol over UDP multicast for market data).

alt text

BTS Spark Server SDK

This API extension point is intended for users that require adding functionality to our Spark Server. Using the Spark Server SDK users can create new order types, new market data sources and soon new execution gateways. In this case users implement the new functionality in C++, producing a shared library that is loaded at run time by our Spark Server.

If you are interested in sending orders or otherwise controlling a spark server remotely please see the BTS Spark Client SDK repo.

In Depth

Getting Started

Docker

We use docker to provide a full dev environment with all the tools users need to develop their own algos.

  1. Setup docker on your preferred OS and pull down the BTS docker
$ docker pull bluetradingsystems/linux_docker
  1. Start the docker container
$ docker run \
       --net host \
       --shm-size=1024m \
       --cap-add=SYS_PTRACE --security-opt seccomp=unconfined \
       -w $PWD \
       -v /home:/home \
       --name bts_sdk \
       -i \
       -t bluetradingsystems/linux_docker \
       /bin/bash

  • Note that the lifetime of docker instance above is tied to the life of the shell instance, so if you exit bash, the container stops running. To start and re-attach to the container, use:
$ docker start --attach -i bts_sdk

Git

Now that you're ssh'd into the docker container, clone the repo into ~/spark_sdk/ and cd to that directory

$ cd
$ git clone https://github.com/BlueTradingSystems/spark_sdk.git
$ cd spark_sdk

Development

Algo Orders typically implement the IAlgoOrder interface to respond to events and use the IAlgoServices interface to create and control child orders as well as manage their own life cycles. See examples/order_plugin_example.cpp for a fully fledged example

Look at examples/md_publisher_example.cpp to look at how to build a custom market data producer

run build_examples.sh to make sure you can build the examples, this will produce an order_plugin_example.so

$ cd ~/spark_sdk/examples
$ ./build_examples.sh

Server Layout

  • Spark servers reside on a port, usually 8000, but your production instance could be on a different $PORT
SDK Package Layout
  • The sdk puts logs in ~/sparkdata/data/logging/log
  • The sdk package has binaries in spark_skd/lib
  • Put plugins in spark_sdk/lib/plugins/ for debugging
Production layout
  • Production Logs are in ~/$PORT/sparkdata/data/logging/log on the host
  • Production binaries are in /sparkbin inside the docker container
  • Production algo plugins should go into the /sparkbin/plugins directory

Recorded market data

Debugging

Copy the order_plugin_example.so to spark_sdk/lib/plugins/ and start the sparkd server process inside gdb

$ cp -f order_plugin_example.so ~/spark_sdk/lib/plugins/
$ cd ~/spark_sdk/lib
$ gdb --args ./sparkd run

In gdb:

(gdb) break order_plugin_example.cpp:366

Running without debugger

$ cp -f order_plugin_example.so ~/spark_sdk/lib/plugins/
$ cd ~/spark_sdk/lib
$ ./sparkd run

Connect to Admin (Risk) UI to set limits or add users

  • Browse to http://location:$PORT
  • Login as bts, no password
  • Add or manage user trading and market data permissions by clicking the Users icon on the left side
  • Add or manage accounts and risk limits by clicking on the Accounts icon on the left side

Other interesting web accessible things

Connect Spark windows client

  • To start a spark client, use the -ConfigServer parameter to point at this new server
> spark.exe -ConfigServer http://location:$PORT/config

Or create a file called spark.conf at %APPDATA%\bts\conf\ that contains:

include: "http://location:$PORT/config/"

Download the Spark windows client

  • Download the bts installer here: