System Performance Testing Environment

Last modified 05 May 2021 11:56 +02:00
Component Role Description

MidPoint

Cluster. Starting with 4 nodes, but number of nodes should be parametrizable later.

PostgreSQL

MidPoint repository

Separate container from all other resources. Reasonable sizing for test size.

PostgreSQL table(s)

Source system

Source data table, filled with data. One table, or several tables. For use with ScriptedJDBC connector Only one source resource for now, but more may come in the future.

Amount of data should be varaible, e.g. we need tests with 100k entries, 1M entries, 5M entries, 10M entries, 50M entries. Let’s start with 100k entries.

OpenLDAP

Target system

Almost empty. Only basic OUs, admin user, ACIs, etc. Sizing for test size, e.g. mdb with appropriate memory setting for millions of entries. suffix: dc=example,dc=com

PostgreSQL table(s)

Target resource

Single database table. For use with DatabaseTable connector. Let’s start with one such resource (one table). More target resource like this will come to testing environment later. The number of such resouces should be easy to change, e.g. run the test with 30 DatabaseTable resources.

Test Scenario Outline

Preparation/preconditions:

  1. Start all resources in clean state (Databases, LDAP)

  2. Start midPoint. Almost empty repo, just resource configuration, administrator, basic roles.

  3. Run sanity checks, e.g. "test connection" on all resources. Schema assertions (optional).

Test scenarios:

  1. Import: Import accounts from source resource. The import should run mappings, create users, provision accounts to target resources.

  2. Empty recon: Run recon task, with the same state of resources and midPoint. Recon should change nothing.

  3. Start livesync task for source resource (timestamp-based). Check sanity (make sure the task runs, token is updated, etc.)

  4. (optional) Small livesync: Initiate small change in source resource (e.g. change a couple of accounts). Make sure that change is picked up by the task and processed correctly.

  5. Big livesync: Initiate big change on the source resource (e.g. changing 10% of all accounts). Make sure the change is picked up by a single livesync run. Make sure that change is processed correctly.