Search code examples
linuxoracle-databaseoracle11gunixodbcotl

what is the optimal number of process that can be handled by unixODBC?

My application used to read data from different sections of large file and update the tables. Each section of the data is associated with a separate table. The update process can contains multiple insert, update and delete statements.

I am trying to split the file into multiple files and running 3, 4 or n instances of the application to achieve parallel execution. I hope this will improve the performance.

Question:

  1. What is the maximum (optimal) allowed process that can handle parallel by unixODBC?

Environment: Red hat Linux 64 bit, C++, unixODBC (32 bit), OTL, Oracle 10,11

Solution

  • unixODBC is just a module which provides APIs compliant to ODBC, not a individual process. To use it, the unixODBC should be linked as a static or dynamic library into the application.

    Many factors can impact the performance: hardwares(CPU, memory), the database design, etc. I think this issue is no related to unixODBC. You can run as many processes which use unixODBC as possible.