The problem is that I need to explain the different sizes of backups that are being made of a database in a plant. Sometimes the difference between the sizes is in negative, even though that there is no data being deleted from the system.
Datum Backupfile-file Size KB Diff 6/1/10 backup201006010100.bak 3355914 7/1/10 backup201007010100.bak 4333367 977453 7/2/10 backup201007020100.bak 4355963 22596 7/3/10 backup201007030100.bak 4380896 24933 7/4/10 backup201007040100.bak 4380404 -492 8/1/10 backup201008010100.bak 4507775 1151861 8/2/10 backup201008020100.bak 4507777 2 8/3/10 backup201008030100.bak 4532492 24715 8/4/10 backup201008040100.bak 4584028 51536
On 7/3/10 and 8/1/10 there was no production. On other days the production is mostly consistent hence the database is expected to have a pretty much linear increase in size, but how is it that the size goes to negative.
In the maintenance plan the the tasks are: Backup Database Task (Type: Full Append Existing) -> Shrink Database (Leave 10% free space)
SQL Server have 2 step process of storing your data. First, your data goes into log file, and it not only data that you inserted, but also the whole list of operations SQL performed on your data. So, if something wrong happens, SQL can ‘replay’ your transactions.
At some point CHECKPOINT happens, ans data gets written into data file. Log files have a tendency to grow and shrink.
During BACKUP, SQL will write data and log files EXACTLY as they look at the point of BACKUP. That’s why you can see that difference in size.