What Is the Impact of Larger Oracle ASM Logs on HVR BFile Capture?

Question

How does increasing Oracle ASM archive or redo log size from 200 MB to 1 GB affect HVR BFile capture?

Environment

• HVR 6
• Source: Oracle ASM
• Capture method: BFile

Answer

With BFile capture, Fivetran doesn't read active online redo logs. We wait for Oracle to archive each log, copy the archived log from ASM to the local staging directory specified by HVR_ORA_BFILE_ASM_PATH, then scan it.

Increasing the log size from 200 MB to 1 GB has the following effects:

1. Oracle must generate approximately five times more redo before switching logs, so archived logs are created less frequently.
2. Capture latency increases. For example, at a redo generation rate of 120 MB per minute, a 200 MB log closes in approximately 1.7 minutes, while a 1 GB log closes in approximately 8.3 minutes. Therefore, Capture jobs may begin processing changes several minutes later.
3. Copy and scan time increases, and more staging space is required. Larger archived logs take longer to copy from ASM and require more temporary disk space. For a two-thread RAC configuration with 1 GB logs, ensure that HVR_ORA_BFILE_ASM_PATH has more than 2 GB of free space.

The following remain unchanged:

• The scan rate in MB per second after we retrieve the archive.
• Sustained catch-up performance.
• Time-based retention policies.
• CPU usage remains broadly unchanged because we perform fewer, larger copy operations.

For near-real-time capture, follow these recommendations:

• Keep archive logs between 100 MB and 512 MB.
• Monitor your staging disc space to ensure it can accommodate larger archive logs.
• Monitor memory usage because we may need to retain uncommitted transactions for longer.

For more information, see Capture From Oracle Using BFile.