Lane/Thread allocation on ATF Test Generator & Cloud Runner

[Oluyinkaoginni]

This article is meant for Automated Test Framework users who use Test Generator and Cloud runner and are looking to increase the speed and accuracy of their test authoring and execution. In this article, I will highlight necessary information needed for additional lane allocation in Test Generator & Cloud runner.

As you should know by now, Test Generator and Cloud runner provides ATF the ability to intelligently author and generate test while also executing them on ServiceNow Cloud infrastructure, this awesome ATF feature has empowered thousands of ServiceNow platform users to generate and execute test with a single click while also ensuring optimal performance and stability of their ServiceNow instances.

In the intricate ecosystem of test automation, efficient resource allocation is paramount. One critical aspect is the allocation of lanes/threads on the ATF Test Generator & Cloud Runner. While the hard limit for allocation  as at the time of publishing this article stands at 20, the nuances lie in the optimal distribution strategy, especially concerning node allocation.

Let's delve into the breakdown:

1. Allocation Limits: The hard limit of 20 lanes per instance sets the upper boundary for allocation. However, below this limit, a standard guideline governs the allocation per node. With each node typically equipped with 8 workers, a maximum of 7 lanes per node is recommended.

2. Practical Allocation Scenarios: For instance, a customer purchasing 4 packs, which totals 22 lanes (including 2 free ones), necessitates at least 4 nodes for efficient utilization.

3. Node-Per-Pack Allocation: A breakdown per pack sheds light on the required node allocation:

   • 1 pack: 7 lanes total, ideally utilizing 1 node, though 2 nodes are typically provisioned for redundancy and load balancing.
   • 2 packs: 12 lanes total, distributed across 2 nodes.
   • 3 packs: 17 lanes total, necessitating 3 nodes for optimal performance.
   • 4 packs: 20 lanes total, ideally requiring 4 nodes. However, the hard limit constraints make selling 4 packs for a single instance impractical until the limit is addressed.

4. Optimization Strategies: To ensure efficient resource utilization, several strategies can be employed:

   • Mutually Exclusive Tests: Implement techniques to isolate mutually exclusive tests, it is advisable to enable sequential execution or smaller group executions to maximize parallelism for non-conflicting tests. In other words, avoid running tests that share the same records.
   • Continuous Monitoring: Regularly monitoring lane utilization and node efficiency enables timely adjustments and prevents resource bottlenecks. This can be monitored via the Browser Orchestration Queue.
   • Dynamic Scaling: Implementing dynamic scaling mechanisms allows for on-demand resource allocation, optimizing performance during peak loads while minimizing costs during idle periods.

5. Future Considerations: Addressing the hard limit constraint stands as a priority for enhancing scalability and flexibility in lane/thread allocation. 

In conclusion, the optimal allocation of lanes/threads on the ATF Test Generator & Cloud Runner requires a nuanced understanding of resource limits, practical scenarios, and optimization strategies. 

Default # of lanes upon installation of Test Generator & Cloud runner	2 Lanes
# of Lanes for a pack	5 Lanes