Selecting a Database for an Algorithmic Trading System
One of the key components of nearly any software system is the database used to persist, retrieve, and analyze data. In this technical post, we talk about the specific characteristics that are important to us in the database serving our Algorithmic Trading Platform. We also take a few contenders for a spin and see how they stack up.
There isn’t a single perfect database that provides support for transactions, time-series data management, and ultra-fast analytics, along with being free or affordable. MemSQL comes close, and combined with its free tier for licensing, is a worthy contender for anyone looking to set up a multi-faceted database for an algorithmic trading system.
We would be remiss if we didn’t give an honorable mention to ClickHouse, which is fast, capable, and free, but just shy of being ready for prime time.
Taking a broad and somewhat simplified view, databases provide the ability to 1) record and manage data (OLTP) and 2) analyze data (OLAP). Most databases are good at one or the other, and even beyond these broad categories, databases have to face competing goals such as performance vs support for ACID transactions.
