Mixed workload test results
A series of mixed workload tests were run (online users plus CMDB CI processing) to determine the hardware requirements for each tier of Helix with Smart IT. During the load test execution, of 2100 concurrent users, 7,500 CIs(90% update and 10% create) were also processed in parallel. Smart IT and BMC Digital Workplace accounted for 85% of the online business transactions run; the remaining 15% were from BMC Helix IT Service Management (ITSM).
Smart IT Restricted Support user had access to 28 End User's companies, were part of 17 Support groups and on an average 120 functional roles . Row level Security was set as Subquery for the forms used during load test . For details when and how to change row level security field to SubQuery refer Smart IT Test Methodology page .
Sizing and scalability
The results show that one Smart IT server CPU can support approximately 240 concurrent users. The results also show that the AR 20.02 System server is the most used tier, and each AR System 20.02 server can scale to approximately 400 concurrent users. Each AR System 20.02 server CPU can support approximately 100 Smart IT users or 150 BMC Helix ITSM users.
The CPU utilization for Helix ITSM , Remedy Single Sign-On, OpenFire, VirtualChat pods was minimal.
Response times and server resource utilization
The bar graphs in this section show end user response times, CPU usage, and memory usage.
The following graph shows the end user browser response times for Smart IT transactions measured manually from the end user client machine with the latency set at 100ms during the 2100-user load test:
The following chart is Smart IT iPad use cases average response times taken from the Silk Performer performance tool during the 2100 user load test.
The following chart is Digital Workplace - Basic use cases average response times taken from iPad Safari browser during the 2100-user load test.
The following chart shows the CPU utilization for each component's container during the 2100-user load test:
Following chart show maximum memory utilized by each component's container during the 2100 user load test.
Prometheus - Monitoring tool, was used to capture CPU, Memory, etc. utilization of application's containers, Grafana tool was used to query and visualize the server metrics.
Perfmon tool was used to capture DB Server usage.