[factormeta]

Likewise. In specific to: >In practice, when used as a data warehouse, SPL does show different performance compared with traditional solutions. For example, in an e-commerce funnel analysis scenario, SPL is nearly 20 times faster than Snowflake even if running on a server with lower configuration; in a computing scenario of NAOC on clustering celestial bodies, the speed of SPL running on a single server is 2000 times faster than that of a cluster composed of a certain top distributed database. There are many similar scenarios, basically, SPL can speed up several times to dozens of times, showing very outstanding performance.

Wow that really sounds amazing? Just wonder how a java based db can out perform Snowflake (a columnar base db). Maybe the original implementation in Snowflake is not optimal? Then again, from personal experience h2 embedded mode significantly faster than plain postgres.

[Judyrabbit]

This post(https://blog.scudata.com/a-major-culprit-in-the-slow-running...) explains why java-based SPL can run much faster than the C++-based database. BTW, SPL also support columnar storage, it can implement columnar storage in a single file. And here is a test report https://blog.scudata.com/spl-computing-performance-test-seri.... It sounds amazing, but it is not mysterious. A lot of low complexity algorithms can not be implemented in SQL, programmer can only expect the optimizer of database. howerver ,when SQL is complex, optimizer would get lost.

[Judyrabbit]

This post(https://blog.scudata.com/how-the-performance-improvement-by-...) also explains the principle of high speed of SPL, more generally