Abstract: Disclosed is a method for improving a database recovery speed using a log data analysis according to a first exemplary embodiment of the present disclosure, including: reading at least one redo log file and loading recovery log data on a storage unit; analyzing the loaded recovery log data and generating a plurality of sub log data groups, the plurality of respective sub log data groups being associated with specific data blocks and the specific data blocks associated with the plurality of respective sub log data groups being different from each other; and generating at least one adjacent log data group based on positional information of the specific data blocks associated with the plurality of respective sub log data groups, each of the at least one adjacent log data group including at least one sub log data group.