The original title is "How SAP HANA is Such a fast Database, but the author's view is that Hana's speed is primarily the result of hardware development, and Hana is not suitable for all applications."
But my focus is on the rest of the conclusion.
Changes in data access patterns
Storage hardware upgrade, from the physical disk to SSD, memory, the corresponding database query mode has also changed.
When a database uses a traditional physical disk, if you read 3 columns in a table with 10 columns, you must also read all 10 columns.
If the table is 100m,100, you must read the entire table to complete the query, which is more serious when multiple table joins are involved.
If your database is stored on a random-access storage such as SSD or memory, you need 3 columns, just read 3 columns.
Elimination of data exchange
The traditional database in disk takes a lot of time to swapping, the data is read into the cache, and then the data is written back to disk, the cycle, endless,:).
The in-memory database eliminates this data exchange (swapping) because all of the data is in-memory, and the only thing that needs to be stored is for data persistence and backup.
In fact, this explains that even if the data is all cache to memory, it is still not as fast as the memory database, whether it is a column database such as Hana, or a row database such as TimesTen.
Conclusion
The column database has the advantage of speed, but not the advantage of omnipotence, although SAP wants Hana as a platform for all, but one Size Does not fit all, the principle of this database, Hana is very suitable for an analytical application such as BW, for the application of OLTP such as ERP , let's wait and see what S/4 Hana does.
Reference
- How SAP Hana is Such a Fast database-scm Focus ' SAP Hana Blog
Why the Hana database is so fast