Update to handling of large records

[Xytam]

Code written by @davidxu8 (David Xu / Weidong Xu)

Oracle internally will handle records dynamically based on record size when passed the length in the setCharacterStream method. By doing this, customers on higher latency connections(>300ms round trip) and with record sizes that may vary above or below the varchar2(4000) limit experience a magnitude improvement in larger records. Additionally, small records on both higher latency and lower latency connections both with and without security and compression options enabled experience roughly the same performance as without this change.

Problem

Connectors handling large record sizes or those that vary above and below the varchar2(4000) limit experience notable performance degradation, especially on connections with higher latency connections due to the way Oracle DB handles large records (which is by disabling message batching). This can be validated by testing by comparing version 5.4.1 of this connector which handled all String fields as CLOBs versus the current version which handles all as varchar2(4000).

Solution

Adding the length field allows the Oracle side to figure out how to treat the record. This does not impact existing setups as it dynamically adds the sizing.

Does this solution apply anywhere else?

• yes
• no

If yes, where?

Test Strategy

Testing done:

• Unit tests
• Integration tests
• System tests
• Manual tests

Release Plan

Reverting or rolling back from this version to a prior version is safe and has no impact on functionality, only performance.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}