To better understand Anonymize’s performance profile and the impact of various configuration options, we conducted a series of tests designed to reflect real-world usage scenarios.

We used the Stack Overflow 2010 10GB database.

Our primary test environment was provisioned with 10 physical cores and 12 logical processors (vCPUs), supported by 32 GB of RAM.

This configuration allowed us to observe how Anonymize responds to a modern, multi-core setup with ample memory resources.

The following sections detail the performance tests we conducted.

Basic Masking

For this test, we set up 2 configurations:

• For the dbo.Users table,
• masking the DisplayName column with the dataset FullNames and 
• masking the Location column with the dataset Cities (US)
• masking the EmailHash column with the dataset EmailAddresses
• For the table dbo.Posts,
• masking the LastActivityDate and LastEditDate columns with the dataset DatesOfBirth (auto varies the date in a range of +/- 90 days)

Processed 4,028,593 rows

Total execution time: 01:07

Classification: 00:01

Mapping: 00:00

Masking: 01:07

Deterministic Masking

For this test, we set up:

• Deterministic Masking synchronizing masked DisplayName values from dbo.Users to dbo.Posts LastEditorDisplayName.

Pairs of values prior to masking - 1,233,688

Pairs of values post masking - 1,233,688

Processed 4,028,593 rows

Total execution time: 01:08

Classification: 00:01

Mapping: 00:00

Masking: 01:07

Increasing the number of columns being masked

For this test, we set up 2 configurations using random in built datasets to mimic using inbuilt datasets across multiple columns:

• For the Users table,
  • masking the DisplayName column with the dataset GivenNames (deterministic)
  • masking the AboutMe column with the dataset StreetAddresses (non-deterministic)
  • masking the Location column with the dataset Countries (deterministic)
  • masking the WebsiteUrl column with the dataset EmptyValue (non-deterministic)
  • masking the EmailHash column with the dataset EmailAddresses (deterministic)
• For the Posts table,
  • masking the LastEditorDisplayName column with the dataset FamilyNames (deterministic)
  • masking the Body column with the dataset Cities (non-deterministic)
  • masking the Tags column with the dataset USStates (non-deterministic)
  • masking the Title column with the dataset StreetAddresses (non-deterministic)

Processed 4,028,593 rows

Total execution time: 03:01

Classification: 00:00

Mapping: 00:00

Masking: 02:59