REORDERING OF DATABASE RECORDS FOR IMPROVED COMPRESSION

• US 20150347426A1
• Filed: 03/11/2015
• Published: 12/03/2015
• Est. Priority Date: 05/27/2014
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Dickie, Garth A., Keller, Jeffrey M.
• Granted Patent: US 9,910,855 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/122 using management policies backup systems G06F11/1446; file migration policies for HSM systems G06F16/185
• CPC Class Codes: G06F 16/1744 using compression, e.g. sparse files
• CPC Class Codes: G06F 16/221 Column-oriented storage; Management thereof
• CPC Class Codes: G06F 16/284 Relational databases
• CPC Class Codes: G06F 7/36 Combined merging and sorting

Abstract

According to embodiments of the present invention, apparatus, systems, methods and computer program products for sorting and compressing an unordered set of data records from a structured database are provided. Fields of the unordered set of data records are prioritized based on an impact of those fields to a compression scheme for column-oriented compression. The unordered set of data records are sorted based on the prioritized field(s) with a greatest impact on the performance metric. Data of the sorted data records are compressed according to a compression scheme. In some embodiments, prioritizing the fields may be based on an anticipated level of usage of data within those fields and/or a cost function associated with a performance metric as well as optimization of compression. A performance metric may include a faster computational time, reduced I/O computation, faster scan time, etc.

Claims

1. A method of compressing an unordered set of data records from a structured database comprising:

prioritizing fields of the unordered set of data records based on an impact of those fields to a performance metric for accessing data stored in a column-oriented compressed database;

sorting the unordered set of data records based on the one or more prioritized field with a greatest impact on the performance metric; and

compressing data of the sorted data records according to a compression scheme.

2. The method of claim 1, wherein prioritizing the fields is based on an anticipated level of usage of data within those fields.

3. The method of claim 1, wherein prioritizing the fields is based on a performance metric as assessed by a cost function, the cost function including minimizing a total amount of time reading and computing on a column as well as the compressed size of data in the column.

4. The method of claim 1, wherein sorting the unordered set of data records further includes:

sorting the unordered set of data records based on a prioritized tuple of sort keys.

5. The method of claim 1, Wherein the unordered set of data records is from a columnar database or a hybrid columnar database.

6. The method of claim 1, wherein the performance metric is determined based upon historical data.

7. The method of claim 1, further comprising:

updating, based upon monitored performance metrics, relative priorities of the various fields to each other;

decompressing a set of ordered compressed records to obtain an unordered set of data records;

sorting the unordered set of data records based on the updated field priorities to obtain a second set of sorted data records; and

compressing data of the second set of sorted data records according to a compression scheme.