Is SQL Fast Enough for Tick Data?
TabbFORUM article by Steven Graves, McObject CEO and co-founder
Published on November 5, 2015
Trading systems have eschewed SQL due to its perceived performance latency and unpredictability, in favor of more labor-intensive, lower-level programming languages (namely q and C/C++) for database operations. But is SQL really too slow for trading systems?
The fact that a SQL DBMS now “owns” nine of the 17 STAC-M3 tests demonstrates that SQL, in and of itself, is not a performance bottleneck. And this stands to reason. SQL (like q) is merely a language; it is the implementation of the language, and characteristics of the underlying DBMS, that determine performance.
So what SQL DBMS characteristics contributed to the record-setting STAC-M3 results?
• Columnar layout
• Distributed Query Processing
Reprinted with permission. Please use the following link to activate the download.
We invite you to evaluate the latest version of McObject’s eXtemeDB for HPC software, ideal for cloud, analytics and Financial systems.