Re: Slow running Delete query for same data volume
From: Mladen Gogala <gogala.mladen_at_gmail.com>
Date: Fri, 2 Apr 2021 14:23:10 -0400
Message-ID: <286c9c7c-1d38-fe70-7032-eb5addd47749_at_gmail.com>
Date: Fri, 2 Apr 2021 14:23:10 -0400
Message-ID: <286c9c7c-1d38-fe70-7032-eb5addd47749_at_gmail.com>
If my memory serves me right, Vn-Vn is "Veeam". Having worked for a backup vendor I know that backups don't necessarily user rman. If Veeam supports so called "synthetic full", then it uses file system backup, not rman. Laurentiu's Observation is rather obvious:
| TotalSeconds | AAS | %This | PROGRAM2 | EVENT2 | FIRST_SEEN | LAST_SEEN |
| 3390 | 4 | 34% | (TNS Vn-Vn) | ON CPU | 3/31/2021 22:24 | 3/31/2021 22:29 |
| 1610 | 1.9 | 16% | (JDBC Thin Client) | ON CPU | 3/31/2021 22:16 | 3/31/2021 22:29 |
| 960 | 1.1 | 10% | (perl) | ON CPU | 3/31/2021 22:16 | 3/31/2021 22:29 |
| 690 | 0.8 | 7% | (perl) | cell multiblock physical read | 3/31/2021 22:16 | 3/31/2021 22:29 |
| 300 | 0.4 | 3% | (DBWn) | db file parallel write | 3/31/2021 22:17 | 3/31/2021 22:29 |
| 300 | 0.4 | 3% | (JDBC Thin Client) | gc cr block busy | 3/31/2021 22:16 | 3/31/2021 22:29 |
| 270 | 0.3 | 3% | (JDBC Thin Client) | cell single block physical read | 3/31/2021 22:18 | 3/31/2021 22:29 |
| 200 | 0.2 | 2% | (Pnnn) | direct path read temp | 3/31/2021 22:16 | 3/31/2021 22:28 |
| 180 | 0.2 | 2% | (LGWR) | log file parallel write | 3/31/2021 22:16 | 3/31/2021 22:27 |
| 170 | 0.2 | 2% | (pmdtm) | ON CPU | 3/31/2021 22:16 | 3/31/2021 22:29 |
| 160 | 0.2 | 2% | (LMSn) | gcs log flush sync | 3/31/2021 22:16 | 3/31/2021 22:26 |
| 130 | 0.2 | 1% | (Pnnn) | ON CPU | 3/31/2021 22:16 | 3/31/2021 22:29 |
| 130 | 0.2 | 1% | (pmdtm) | cell single block physical read | 3/31/2021 22:20 | 3/31/2021 22:29 |
| 110 | 0.1 | 1% | (LMSn) | ON CPU | 3/31/2021 22:17 | 3/31/2021 22:29 |
| 90 | 0.1 | 1% | (oracle) | ON CPU | 3/31/2021 22:20 | 3/31/2021 22:28 |
| 90 | 0.1 | 1% | (oracle) | direct path read temp | 3/31/2021 22:16 | 3/31/2021 22:29 |
| 80 | 0.1 | 1% | (TNS Vn-Vn) | Backup: MML commit backup piece | 3/31/2021 22:27 | 3/31/2021 22:28 |
| 60 | 0.1 | 1% | (TTnn) | ON CPU | 3/31/2021 22:16 | 3/31/2021 22:29 |
| 50 | 0.1 | 1% | (ARCn) | log file sequential read | 3/31/2021 22:17 | 3/31/2021 22:24 |
| 50 | 0.1 | 1% | (pmdtm) | direct path read temp | 3/31/2021 22:22 | 3/31/2021 22:23 |
So, Veam is using rman (Backup MML).
The consumption of CPU probably comes from compression,
deduplication or the combination of both.
On 4/2/21 6:46 AM, Pap wrote:
Thank you. So the program (TNS vn-vn) is the one for backup. But, I am not able to see any DB incremental backup run from v$rman_backup_job_detail in this database during that time. And we are using ZDLRA , so all are incremental backup only. I see there was an archivelog backup running between ~10:20PM till 10:30PM. Can that impact to such extent?
Why are you using Veeam if you have ZLDR appliance? Zero Loss
appliance in itself should be protection enough.
-- Mladen Gogala Database Consultant Tel: (347) 321-1217 https://dbwhisperer.wordpress.com-- http://www.freelists.org/webpage/oracle-l Received on Fri Apr 02 2021 - 20:23:10 CEST
