Sizing Guidance for PRA Workloads

This document offers guidance to users to determine what hardware configuration is needed to host the PRA workload depending on the number of concurrent PRA session that engine needs to host.

Size testing was performed on Platform Engine hosts equipped with an Intel(R) Xeon(R) Gold 6254 CPU @ 3.10GHz 3.09 GHz processor, 64-bit operating system, x64-based processor and with 6GB of RAM.

Sessions were tested on a site with a single Platform engine with PRA capabilities. No other capabilities were added to the engine.

Multiple concurrent sessions were opened, kept running for 30 minutes and then closed. The following actions were performed on the remote target machines after connection was established:

  1. SSH: Login to a remote Linux server and run “top”.

  2. RDP: Login to a remote Windows server and run a batch script that produces output similar to that of a typical command-shell script.

Session recording was disabled in both test cases.

Session Type # of PRA Workloads # of Concurrent Sessions CPU Usage % Memory Usage in MB
SSH 1 200

0.357 avg

(8.921 max)

13.21 avg

(13.21 max)
RDP 1 100

0.706 avg

(24.297 max)

13.54 avg

(16.21 max)

Your actual capacity may vary depending on the specific activities that are being carried out on the remote target machine. For example, viewing a video clip on the remote machine on RDP will put very different load on PRA as compared to running vim to edit a text file on a remote SSH session. Performance test data also indicates that PRA performance is more dependent on memory and bandwidth than CPU speed or power.

The network latency between the end-user’s browser and the Delinea Platform, or the latency between the Platform Engine and the target server also affects the user-experience.

We recommend the following steps when experiencing PRA performance issues that may be associated with the Engine:

  1. Use system monitoring tools to determine whether any specific resource like memory, network bandwidth, CPU utilization is running at unacceptably high levels. If possible, increase the corresponding hardware or virtualization resources.

  2. Deploy more engines to the parent site.

  3. Check the network latency between:

    1. The Engine and the Delinea Platform tenant.

    2. The Engine and target machines

  4. Contact Delinea Support.