Test setup and benchmark results

Four different extract rules were used. The Simple extract rule has twelve properties of mixed types (Text, Decimal, Double, Date, DateTime, and others).

Complex Level 1 extract rules have the same properties as Simple with the addition of the Page List and Page Group properties. Each embedded page has twelve properties.

Complex Level 2 has the same layout as Complex Level 1 plus one additional layer of Page Lists and Page Groups.

Complex Level 4 has the same layout as Complex Level 2 plus two layers of Page Lists and Page Groups.

Each extract rule was executed three times, each with 100000, 200000, 300000, 400000 and 500000 records (work objects). The chart below presents a summary of the extract rules.

Data structure

Data structure

System software

Benchmark results

The following graph shows the performance of a Comma Separated Values (CSV) extract for each set of records. For example, 500,000 Complex Level 4 records were extracted into a CVS file in 4 hours and 48 minutes, or about 29 records per second.

The following graph shows the performance of an XML extract for each set of records. For example, 500,000 Complex Level 4 records were extracted into an XML file in 3 hours, or about 46 records per second.

The following graph shows the performance of a Database schema extract for each set of records. For example, 500,000 Complex Level 4 records were extracted into a database schema file in 5 hours, or about 28 records per second.

The following graph shows the comparison between each type of extract for a Complex Level 4 data structure.

The following graph shows the amount of memory that was used during the extract.