In-memory database vendor McObject is new to me, though recent Indian customer press releases – here and here – piqued my interest. It seems the company has been around for a good few years, has around 300 customers, including a few in the financial markets, and has built some interesting technology, called eXtremeDB.
The secret sauce of eXtremeDB is that it was created from scratch for embedded applications, and has found uses in everything from satellite TV boxes and mp3 players to telecoms networks and Tornado fighter jets. Oh, SunGard is a customer too, as is China’s Dalian Commodity Exchange. Proprietary trading firms in New York City and Chicago – who can’t be named – also use it for bond and algo trading, portfolio management and in one case for a matching engine.
Because it was designed for embedded applications, the code has been written to run on slow processors with limited memory. The evaluation code – which runs in process with the target app – takes up just 150KB, and that can be reduced to as little as 50KB. By definition, it’s fast. It doesn’t have enough code to be slow!
It’s also pretty functional. Highlights being SQL, ODBC APIs, 32-bit and 64-bit versions, multi-platform, low overhead indexing, relational or object storage as native, persistence, C, C++, Java, .Net interfaces. Oh, and multiple-dimensional-indexes – consider searching for equities by symbol and by price, and by trading volume – all using just one index.
The recent wins in India include the National Stock Exchange, which needed to get its order processing time to less than a millisecond, despite implementing a number of pre-trade risk checks. eXtremeDB is the database supporting those risk checks.
The second win is at software vendor Financial Technologies of India, which will use eXtremeDB is its Windows-Server-based DOME (Distributed Order Management Engine) trading platform.