Parallels RAS Performance Monitor 

Parallels RAS Performance Monitor is an optional component download msi package. Parallels RAS Performance Monitor provides browser-based dashboard to help administrators to analyze a Parallels RAS farm deployment bottlenecks and resource usage.

Overview

Components

Parallels RAS Performance Monitor consists of the following components:

• InfluxDB database — a database for storage of system performance data.

• Grafana dashboard — a browser-based dashboard which provides a visual display of performance metrics.

• Telegraf service — a service that collects performance data on a server where it is installed. The service is installed automatically when you add a server to a Parallels RAS farm and install a corresponding RAS Agent on it (e.g. RAS Secure Client Gateway Agent, RD Session Host Agent, Remote PC Agent, etc.).

How it works

The Telegraf service is stopped by default, so it doesn't collect any data. To start the service on all Windows-based servers in a farm, the performance monitoring functionality must be configured and enabled in the Parallels RAS Console. Once enabled, the Telegraf service begins collecting a predefined set of performance counters at a fixed time interval (10 seconds). It then sends the collected data to the InfluxDB database for storage using port TCP 8086. To view performance metrics, the Parallels RAS administrator uses the dashboard (Grafana, port TCP 3000), which displays the visual representation of performance counters in real time.

Collected counters

Host performance

1. CPU usage (win_cpu series) : "\Processor(_Total)\". Counters: "% Idle Time", "% Interrupt Time", "% Privileged Time", "% User Time", "% Processor Time".
2. Physical memory usage (win_mem series): "\Memory\". Counters: "Available Bytes","Cache Faults/sec","Demand Zero Faults/sec","Page Faults/sec","Pages/sec","Transition Faults/sec","Pool Non-paged Bytes","Pool Paged Bytes".
3. Network interface I/O traffic (win_net series): "\Network interface(*)\". Counters: "Bytes Total/sec","Packets/sec","Current Bandwidth","Output Queue Length".
4. Disk I/O (win_disk series): "\LogicalDisk(*)\". Counters: "% Idle Time", "% Disk Time","% Disk Read Time", "% Disk Write Time", "% User Time", "Current Disk Queue Length".
5. Context switches and system calls (win_system series).

Application performance

1. Number of active sessions, number of disconnected sessions (reported by TS component)

Host performance impact

According to our lab tests, collecting of CPU counters affects the host performance (LoginVSI max value is dropped by 2 connections in average) even if they collected once in 10 seconds. Collecting of other counters doesn't have any performance impact.

Parallels Support

Parallels Support encourages you to deploy and use the Parallels RAS Performance Monitor. When you encounter a performance issue, send a problem report to Parallels Support for analysis. The problem report will include the perf.dat file retrieved from the InfluxDB database and added to the zipped problem report file. Providing additional information will help with the resolution of an issue.