The objective of this benchmark is to characterize the performance and scalability aspect of a deployed AR System ITSM stack under various field conditions. For a realistic simulation, the benchmark workload must include both online and batch workloads. For details about the workload, see Definition of performance benchmark use cases.
A supporting environment was built and IT infrastructure was used in this benchmark to depict a typical deployment architecture; in particular, the HA requirement of many critical deployments.
The workload simulations include the following:
- A mixed workload simulating ITSM, SRM, and Reporting users with Atrium CMDB processes running in continuous mode
- A standalone Remedy Single Sign-On workload
- A standalone Virtual Chat workload
- A mixed workload simulating virtual chat users, ITSM, SRM, and Reporting users with Atrium CMDB processes running in continuous mode on Single mid tier
The supporting lab environment includes the following:
- An F5 load balancer
- The AR mid tier web application deployed as a cluster on multiple tomcat instances
- Multiple instances (HA) of the Remedy Single Sign-On (RSSO) web application with each instance deployed in the same tomcat as the AR mid tier web application
- Exactly one Remedy Smart Reporting Server
- Exactly 2 AR System servers configured as an AR Server Group
- For Virtual Chat workload the AR mid tier web application was deployed on single tomcat instance
Following are the key takeaways of the benchmark results:
- For the 600 concurrent users mixed (online and batch) workload benchmark, the average end-user response time per use case is comparable to the Remedy 9.0 release.
The Remedy ITSM suite guideline for hardware sizing is consistent and identical to Remedy ITSM Suite.
When the new Always On Logging (AOL) feature is on, an additional 3% to 10% CPU utilization is observed on the AR server and this new feature is configured to be ON out-of-the-box. However, this option can be turned OFF if not needed.
- The AR mid tier deployed as a cluster scales horizontally and vertically.