FIX: Slow database recovery in SQL Server 2016 due to large log when you use In-Memory OLTP on a high-end computer

Applies to: SQL Server 2016 DeveloperSQL Server 2016 EnterpriseSQL Server 2016 Enterprise Core


In SQL Server 2016, if SQL Server detects itself running on a high-end machine, and memory-optimized tables are used, SQL Server will enter into the Large Checkpoint mode. Additionally, the following informational message is logged in the SQL Server error log:

In-Memory OLTP initialized on highend machine.

In the Large Checkpoint mode, an automatic checkpoint is taken when the transaction log file is larger than 12 gigabytes (GB), which is eight times larger than the normal mode. However, database recovery may be very slow in this mode because of the large log (may be as large as 12 GB).


After you install this update, automatic checkpoints for In-Memory databases are the same as previous releases instead of 12 GB.

The issue was first fixed in the following cumulative update of SQL Server:


Microsoft has confirmed that this is a problem in the Microsoft products that are listed in the "Applies to" section.


Learn about the terminology that Microsoft uses to describe software updates.