SQL Server Managed backup is a new feature introduced in SQL Server 2014, this feature is simply outstanding – you don’t have to worry about the backup strategy anymore, this feature manages and automates the backup process.
This feature uses sort of artificial intelligence – if you see at current scheduling features they just keep on taking backups at a specified time interval whereas, in this feature it checks couple of things and based on the situation it decides whether a backup needs to be initiated for a Database or not. This Feature is based on Retention period selected for backups and transaction workload on the database and it also provides Point-in-time recovery. You want to know how? Keep reading… you’ll get your answers soon.
You can enable this feature at two levels Instance as well as Database level. It doesn’t really matter if your SQL Server is running On-Premise or on Cloud. SQL Server Managed Backup to Microsoft (Windows) Azure is in-fact recommended for SQL Server running on a VM in Cloud.
The Retention period is a value which determines what backup files to keep and what needs to be deleted in the Azure Storage so that the underlying DB can be recovered to a point in time within the time frame specified. The valid retention period values are from 1 till 30 days. The backup_finish_date of the backup is used to determine and match the time specified by the retention period settings.
Naming convention it follows for the backup and container
The backup files are named using a combination of first 40 characters of the database name, the GUID of DB without “-“& the timestamp; all these values are separated using Underscore to make it readable. The .bak and .log extensions are used for Full and Log Backups respectively. In case of Availability Group databases the Availability group database GUID is added after the first 40 characters of the DB name, you can get this GUID value from group_database_id column of Sys.databases.
The same way Backup container of (Windows) Microsoft Azure also get named using the SQL Instance name and in the case of Availability databases it gets named using GUID of Availability group.
Conditions that trigger a backup
As we discussed earlier also SQL Server managed backup doesn’t initiate a backup merely on the basis of schedule which has been fixed but takes a calculative decision whether a backup is really required for the given DB or not. It decides based on the below mentioned conditions:
A. Full backup gets triggered:
1. At the time when Managed Backup is configured on DB or when it is configured on the complete instance with default settings.
2. If the log generation since last full backup is greater than or equal to 1 GB in size.
3. If maximum time interval of one week has passed since the last full backup. (it tries to take at least one full back up every week)
4. When the log chain is broken; it checks periodically whether the log chain is intact or not by comparing first and last LSN of backup files, if it finds there is a gap it initiates a full DB Backup. There could be n number of reasons for log chain breaks, most common would be:
a. Switching recovery model to simple.
b. Taking back up through any other job\manual process.
c. Deleting some backup files
d. Accidental overwrites of backup files.
e. Truncating the log.
B. Transaction Log backup gets triggered:
1. When the Transaction Log space used is equal to or greater than 5 MB.
2. The maximum time interval of 2 hours has passed since the last log backup.
3. If no log backup history is found; for example when you enable Managed backup for the first time for the respective DB.
4. Anytime the TLog backup is lagging behind the full backup, it always tries to keep the log chain ahead of the Full backup. In simple words after every full backup it will initiate one TLog backup.
Sarabpreet Singh Anand