I work for a county government and we are trying to determine the best way forward with an issue.
We have two employees who edit parcel data full time directly in the SDE database. This makes the transaction log in SQL fill up really quick. Recently it has caused errors because it got full. Is this because they are editing the SDE data directly? Eighty hours a week of edits seems like a lot of transactions to log (if I understand things right).
So we are talking about switching to replication so they would edit a replica of the data and then it would synchronize at the end of the day (either manually or by script). Would this decrease the transaction log? In other words, would there be one transaction for the synchronizing OR would the synchronization still maintain all their transactions in the replica and log all of them during synchronization?
The people who set this system up are no longer here and those of us left have only rudimentary knowledge of SQL and how it all works.
Thank you for any advice.
This is really a question about how much "coverage" you want to have in regards to disaster recovery of your SQL Server database. There are different recovery methods in SQL Server that can be used. The transaction log is not an Esri controlled piece of the SQL Server.
Here are some good links to start looking at managing the transaction log files:
Hope these help point you in the correct direction.