Before going to market, verifying and validating the performance metrics to meet either business goals or end-user expectations is a constant challenge for any software product, especially after introducing new features.
That’s why the Workload Automation performance team focused on performance, scalability, and reliability of the new features delivered in Workload Scheduler V10.1.0.0 and V10.1.0.1, in addition to consolidating performance improvements observed with the previous releases (V9.4 and V9.5).
New functionalities added with 10.1 and 10.1 FP1
Workload Automation 10.1 and Fix Pack1 went to market in 2022.
If you’re curious about the new features introduced, look at the summary of enhancements in the HCL Workload Automation documentation:
The changeover introduced several adjustments to the product infrastructure and functional capabilities. Performance testers focused on some of these changes to understand whether the changes might affect the overall performance and scalability of the product by comparing them to the previous releases’ results.
Let’s look at the 4 performance test focus areas:
- Automatic encryption at rest for the critical product files: A brand-new installation of the server that automatically encrypts the key files (such as the symphony file, messages queues, and the useropts file) by using AES-256 or AES-128 cryptography. After having installed 10.1 FP1, you can apply the encryption, also to the pre-existing environments.
- New Workload Designer application in the Dynamic Workload Console: To simplify the user experience and create a more responsive, fast, and fluid user interface, we introduced an innovative infrastructure and design.
- Automatic SSL configuration for fault-tolerant agents (encryption in motion): To ensure encryption in motion for all data moving within customer environments, you can configure IBM WebSphere Application Server Liberty Base and fault-tolerant agents in SSL mode and provide custom certificates at installation time by setting the sslkeysfolder and sslpassword parameters in the installation command line.
- Support for file transfers via the Workstation-to-Workstation internal protocol: Only for master domain managers (MDM in this publication) and agents at version 10.1 or higher, it’s possible to transfer files to and from agents connected to the same master domain manager without specifying the address of the workstations involved in the file transfer. It is enough to define the workstations from the File Transfer integration.
Performance Test Results Overview
The main scope of the Workload Automation Performance assessment is to run the performance test in a test environment based on VMWare ESX – Linux x86, observe and analyze the results of the new version of the product (V10.1.0.1) and compare them to the previous version metrics (see Workload Scheduler 188.8.131.52 performance report).
Let’s go through the summary of the performance results:
- The throughput analysis confirms the performance and scalability levels assured in previous releases.
Even during huge job submission peaks, the Workload Automation user does not see substantial delay in the jobs and job streams status update on the Dynamic Workload Console. The figure below shows the CPU resource utilization trend on the MDM machine with respect to the outcoming throughput for the dynamic agent and fault tolerant agent scheduling. It is interesting to underline how the MDM CPU footprint for 10.1 and 9.5 product versions are similar, even if 10.1 version has added new functionalities. This guarantees greater security enforcements.
Figure 1 Total average CPU utilization at Master Domain Manager vs different workload
- After enabling the new automatic encryption at rest and encryption in motion features, there are minimal performance impacts on product throughput capabilities and master domain manager CPU utilization during jobs submission peak. These performance impacts are balanced by enabling a more secure configuration for the product.
- The scalability and the performance of the Dynamic Workload Console confirm the same level of quality assured in previous releases.
Figure 2 Average response time over time for all pages
For a test environment with around 43,000 job streams and 500,000 jobs in plan each day, the maximum response times have been observed for the monitoring jobs’ direct quer–with average response times within 5 seconds.
- The new Workstation-to-Workstation file transfer internal protocol guarantees a faster file transfer scenario when compared with the previous implementation. The improvements measured using the new protocol, in relation to the file transfer execution time, are in the range between 27% and 43%.
Figure 3 File Transfer scenario implementation using Workstation-to-Workstation protocol
This is, of course, only an excerpt from the unabridged performance report. If you have a further interest in the topics discussed, see the full report version 10.1.0.1, HERE