|
1. Real-World Workloads Go Untested:
Load tests hammer large volumes of repetitious test data using a small number of test cases. This leads to a poor transactional mix, neglecting the vast combinations of logical journeys and data that users might exercise. Performance under most production-like workloads remains untested.
|
1. Maximal Transactional Mixs:
Automated coverage algorithms identify every combination of data and end-point reflected in quick-to-build flowcharts. This rapidly creates the smallest number of tests required to simulate all production-like workloads, achieving the greatest transactional mix possible in short iterations.
|
|
2. Slow and Complex Test Scripting:
Automated test execution is necessary to reflect large volumes of concurrent users, but writing Load Test scripts manually is slow and labour-intensive. It is furthermore complex as tests must be parameterised for concurrency, ramp-up time, and more.
|
2. Automated Load Test Automation:
Test Modeller generates Load Test scripts for a range of frameworks, including Open Text (Micro Focus) LoadRunner, JMeter, and Taurus. Code from existing frameworks can be parsed and imported, building parameterised tests with a simple, visual builder.
|
|
3. Test Data Leaves Systems Exposed:
Realistic Load testing requires a vast number of data combinations, reflecting user activity, message data and API calls. Load tests burn through this data rapidly, and provisioning copies of production cannot keep up. Production data is also low-variety, hitting just a fraction of system end-points, while cross-team constraints prevent parallel testing.
|
3. Test Data for Every Workload:
Complete test data is found or made as tests are generated. Synthetic test data functions define data dynamically at the model level, and resolve “just in time” during test creation. Repeatable data “finds” identify data in back-end systems, automatically creating any missing data. Data is accessible instantly and in parallel, quickly simulating every workload.
|
|
4. Hard-to-Define Message Data:
Load testing with APIs further requires JSON messages with which to inject the test data. Testers must formulate numerous Payloads with complex logic, including embedded structures and arrays.
|
4. Quick and Simple Message Creation:
A simple and visual message builder designs JSON Messages, selecting from available test data variables. A drag-and-drop approach orders the payloads, quickly designing logically complex JSON message data.
|
|
5. Test Environments Undermine Agility:
Accurate Load testing for distributed systems requires access to every component in that system, as well as environments that reflect every application, operating system and device used in production. Many of these components might be unfinished or in use by another team, while slow provisioning of costly and out-of-date environments prevents parallelized testing.
|
5. On Demand Test Environments:
Automated Load Tests and data generated in Test Modeller are executed automatically using a range of open source and commercial frameworks. Integrations with platforms like Sauce Labs provide on demand access to test environments. These include multiple browsers, operating systems and devices, while virtual services simulate any unavailable components.
|
|
6. Complex End-to-End Testing:
The number of possible combinations of data and user activity contained in a system grows rapidly as new components are introduced, including complex chains of API calls. There is no way to identify every distinct Workload manually, let alone create the Load tests to mimic them.
|
6. Load Tests for Complex Systems:
Every model created in Test Modeller becomes a re-usable subflow and can be dragged-and-dropped to create end-to-end tests. This rapidly chains components together, generating covered Load Tests for chains of API and database calls, as well as for linked-up UI screens.
|
|
7. Maintenance Lags Behind Sprints:
Test Scripts, test data and environments must be checked and updated manually each time the system changes. This is highly time-consuming, forcing testers to either execute invalid Load tests or leave Workloads untested.
|
7. Rigorous Load Testing In-Sprint:
Load test maintenance is as quick and simple as updating the central flowchart models, re-generating the test cases, scripts and data. Rigorous Load Testing occurs within an iteration, keeping up with even short sprint cycles.
|
Model-Based Load Testing
