Lyrebird · Realistic synthetic test databases, generated fully offline

SQL Server · AdventureWorksLT

dbo.BuildVersion

Source rows
1
Synthetic rows
1
Broken foreign keys
No foreign keys
Rows equal to a source row
Keys only

Rows

Both databases, first rows by primary key, in one table: scrolling sideways moves them together. Rows at the same position are unrelated, unless the table holds kept reference data. Long values are cut at 60 characters.

SystemInformationIDDatabase VersionVersionDateModifiedDate
Source database · 1 of 1 rows
110.50.91013.002009-10-13 00:00:002009-10-13 00:00:00
Synthetic database · 1 of 1 rows
113.01.78117.052009-10-13 00:00:002009-10-13 00:00:00

Column by column

Each figure is computed from all rows of both databases in the same way. S is the source, Y the synthetic copy. Difference runs from 0 (same distribution) to 1 (no overlap): Kolmogorov-Smirnov for numbers and times, total variation for columns with few values. Personal columns are meant to differ, so they show how many values also appear in the source instead. Lyrebird draws names from built-in lists, not from the source, so common names can match.

Column Model read it as How it was generated Empty Distinct Range or length Most common Difference
SystemInformationIDtinyint record key Numbers from the measured histogram S 0%
Y 0%
S 1
Y 1
S 1 → 1
Y 1 → 1
S 1 100%
Y 1 100%
0 KS
Database Versionnvarchar(25) sequence number New text in the measured format S 0%
Y 0%
S 1
Y 1
S 14 chars avg
Y 14 chars avg
S 10.50.91013.00 100%
Y 13.01.78117.05 100%
1 TVD
VersionDatedatetime lifecycle timestamp Times from the measured distribution S 0%
Y 0%
S 1
Y 1
S 2009-10-13 00:00:00 → 2009-10-13 00:00:00
Y 2009-10-13 00:00:00 → 2009-10-13 00:00:00
S 2009-10-13 00:00:00 100%
Y 2009-10-13 00:00:00 100%
0 KS
ModifiedDatedatetime lifecycle timestamp Times from the measured distribution S 0%
Y 0%
S 1
Y 1
S 2009-10-13 00:00:00 → 2009-10-13 00:00:00
Y 2009-10-13 00:00:00 → 2009-10-13 00:00:00
S 2009-10-13 00:00:00 100%
Y 2009-10-13 00:00:00 100%
0 KS