RE: Synchronous Redo Transport Maximum Availability || Datagaurd

What an excellent summary! (My SWAG on that is about 200 manual pages plus experience summarized accurately with respect to trade-offs in one paragraph.)  

I’m going to add two little bits:  

1. IF one of your destinations is in the same computer room as the primary on a dedicated high speed link, the SYNC penalty is still there but is minimized.
2. IF at least one of your redo group members and one of your archive destinations is a device has Axxana type remote access despite a “campus disaster” then the penalty to DEFER with respect to recovery is minimized.

mwf  

From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Jon Crisler Sent: Friday, February 10, 2023 1:03 AM
To: chrishna0007_at_gmail.com
Cc: Oracle L
Subject: Re: Synchronous Redo Transport Maximum Availability || Datagaurd  

If the Dataguard config was done with the DG Broker, then it would be easier to change the protection mode via the broker as described at https://docs.oracle.com/database/121/DGBKR/cli.htm#GUID-82319941-58E8-4672-8609-7CC496D3DC29

Anytime the redo transport is using SYNC, it will have some impact when there is a data change involved. except in the case of MAXAVAILABILITY when the standby db is down. Just run the DG change during off-hours to be safe. If you DEFER the destination then the redo logs will not ship. When using SYNC, the transaction has to be committed on the primary and at least one standby, so every insert/update/delete will have a wait component while the data is being shipped and applied. You do not mention Active Dataguard, or network speed and latency, or use of FARSYNC, and any of these affect the outcome. But in normal conditions the SYNC has a penalty . Now, if you are doing the DEFER on a daily basis, I think you would be better off just using MAXPERFORMANCE 100% of the time. If the DEFER is only for occasional use and you are not using MAXPROTECTION then I do not have a firm opinion on the impact, as it has been a while since I tested that combination.  

Sorry if this is not a clear opinion , but there are different tolerances for DG lag, protection modes and possibility of data loss.  

On Thu, Feb 9, 2023 at 5:05 AM Krishnaprasad Yadav <chrishna0007_at_gmail.com> wrote:

Dear Experts ,  

Customer has query where he have implemented Max availability as Protection mode by setting LGWR SYNC NOAFFRIM/AFFRIM for one of their critical system to their DR location for minimal data loss , they want to know if there is spike seen in database because of it and they want to go back and disable fast sync /AFFRIM , then do they just need to change DR archive dest location to defer or they too need to downgrade the database mode from max availability to max performance.  

Concern here is that they don’t want to use any alter database command in its peak time and considering deferring of DR dest they feel is a safer option , however we are not clear what would be the implication if we simply defer the dest without changing the mode of database from Max Avail. To Max performance  

Database version is 19.15 , OS : AIX 7.2  

Request you to share some light on the same .  

Regards,

Krishna

--
http://www.freelists.org/webpage/oracle-l