Getting Started with Endpoint Agents

As applications now run in the cloud and workers are more distributed, you can use ThousandEyes end-user monitoring with the Endpoint Agent to gain visibility into your employees' experience of SaaS and internally hosted applications, as well as the underlying wireless LAN, WAN, internet connectivity, and system health.

Endpoint Agents in the Broader ThousandEyes Ecosystem

Before learning how an Endpoint Agent works, let's first explore how the Endpoint Agent relates to the other ThousandEyes vantage points. First, compare the context-to-detail ratios of the different ThousandEyes products:
  • Internet Insights gives you macro-level visibility into outages that may affect you, using the collective intelligence of ThousandEyes. Answering questions like “Is my organization the only one having this issue”, “What is the blast radius of this outage”, or “Which set of providers has the most outages in a certain region”.
  • Cloud and Enterprise Agents provide deep insights in your organization's digital experience. These always-on vantage points, the information they give, and the alerts they trigger are useful for identifying and isolating outages and performance degradations for the applications and locations tested.
  • The Endpoint Agent provides the most detail, augmenting the path and application metrics with local browser, machine, and network data (including wireless).
The Endpoint Agent is a proactive tool. We do not recommend that you use the Endpoint Agent in a reactive manner by installing an agent ad-hoc when a user reports issues.
The Endpoint Agent monitors significant applications and network health, and silently connects to them in the background while users are actively working. This functionality enables support teams to retrospectively review the exact time an issue took place, which promotes a proactive response. Conversely, waiting for the error to resurface and then installing the agent is counterproductive.

Choosing the Right Targets for Agent Installation

The first task in your Endpoint Agent deployment is to plan where to install agents. The most successful customers install the Endpoint Agent on a broad set of users, preferably every machine in the organization. This broad distribution allows you to troubleshoot the most issues and provides important trending information on application performance.
Best Practice: Bulk Installation
A successful ThousandEyes deployment shouldn’t depend on the manual installation of agents. Involve your organization's desktop team early in the planning process so your organization can do an automated installation.
Before you deploy the automated installation, make sure that the browser plugin is included in the installation package. For guidance on how to create a bulk install, see Install Endpoint Agents for Windows via Group Policy.

Leveraging Endpoint Agent Labels

Once your Endpoint Agents are installed, use labels to assign tests to agents, and to make dashboards based on these labels.
Unlike a Cloud or Enterprise Agent, an Endpoint Agent is dynamic by nature. The owner of a laptop could be in the office with wired Ethernet, or working from a coffee shop on wireless. You will likely want to run different tests depending on the particular device context: for example, you should only test internal applications for endpoints that are connected to the VPN or internal network.
For detailed information on creating labels, see Configure Endpoint Agent Labels.

Start with Common Labels

Any particular label configuration will be organization-specific, but there are some good places to start. Consider creating specific labels around the following categories:
  • Connection-based: wired, wireless, and VPN connections.
  • Location-based: relevant parts of the world, the country you are located in, building or floor, etc.
  • Network-based: office networks, or important ISPs where remote workers are working from, to gain insights into performance differences between different groups of remote workers.
  • Agent-specific: only for specific agents based on organizational groups, rather than being dynamically assigned based on device conditions.

Example

Let's say you want to create connection-based labels.
To create the labels for wired and wireless, the process is straightforward: When you configure your labels, set the Connection field to “Ethernet” or “Wireless”.
To create the label for VPN-connected Endpoint Agents, there are multiple ways of selecting VPN users. You can base the selection on VPN vendor, gateway address, client network, or client IP. The image below shows selection based on the VPN client network, adding the VPN label to every agent that is part of any VPN network.
Navigate to Endpoint Agents > Agent Settings > Agent Labels > Add New Label:

Configuring Endpoint Agent Tests

After installing the agent, you need to configure the right set of tests. The ThousandEyes Endpoint Agent provides you with the following different types of tests:
  • Real user tests dynamically capture website, network, and agent details using the ThousandEyes Endpoint Agent browser plug-in. Based on the session and agent statistics, an experience score is computed to help you better understand the user's interactions.
  • Synthetic Tests: A synthetic test template combines scheduled and dynamic tests to monitor a specific application. All the major applications have a defined synthetic test template, and you can also create a custom template for applications that do not have a predefined template.
    • Scheduled tests are HTTP server or network tests that are run from Endpoint Agents at regularly scheduled intervals. Hence, you get a continuous baseline from the end user's vantage point.
    • Dynamic tests automatically identify and test remote targets for collaboration applications, based on observed network connections to dynamic remote servers.

Real User Tests

Real user tests are recorded automatically when the user visits a website in the monitored domain set.
A monitored domain set consists of domains you want to gather end-user performance metrics about. The best practice is using business-relevant domains for tools and sites your end users access.
To configure real user tests:
  1. 1.
    Navigate to the Endpoint Agents > Test Settings > Real User Tests tab.
  2. 2.
    Click the Add New Monitored Domain Set button.
  3. 3.
    Configure the basic settings for the session.
  4. 4.
    Assign the agents that run the real user test for this Monitored Domain Set. There are three options available:
  • All Agents: The test is assigned to all available agents. This is selected by default.
  • Specific Agents: The test is assigned to user-defined agents. You can select the agents from the drop-down or the search function.
  • Agent Labels: The agent is assigned based on pre-configured labels. See Configure Endpoint Agent Labels for more information.
  1. 5.
    Click Add New Monitored Domain Set at the bottom to save the domain, or click Cancel to discard the changes.
In the example image above, a new monitored domain set (called "BusinessDomains”) is created, with three domains added: ThousandEyes.com, Cisco.com, and CiscoLive.com.
A monitored domain set includes all of its subdomains. For example, adding microsoft.com to a monitored domain set automatically includes office.microsoft.com in real user tests.
For more information on real user tests, see Real User Tests View.

Scheduled Tests

Scheduled tests monitor the availability and performance of web applications or network targets. For example, use them to test corporate VPN gateways or critical business applications.
To configure a scheduled test, go to Endpoint Agents > Test Settings. You can configure a web-based or network test, and set a test interval ranging from every minute to once every hour.
A powerful feature of scheduled tests is the ability to test based on the Endpoint Agent label (as shown in the image below).
The example test in the image above executes on agents that have the VPN label. This ensures that the test is not run for users who do not have access to the corporate network. At the bottom of the Add New Test dialog, you can find the agents that currently match this label.
When getting started with scheduled tests for Endpoint Agents, consider the sites and applications your users rely on, and create tests that provide the best visibility.

Dynamic Tests

Dynamic Tests enable the Endpoint Agent to monitor and identify network connections between an end-user’s application and the destination node (host server). This removes the ambiguity around knowing what applications require testing. Dynamic tests monitor applications like collaboration and communication tools, such as Webex. As you get started with your Endpoint Agents, we recommend enabling dynamic tests for the collaboration apps your organization uses.
The Endpoint Agent watches for remote connections from specific applications on the user’s machine, and automatically runs tests toward those remote targets. For example, a dynamic test for Webex will detect when a user joins a Webex meeting and will execute tests from the Endpoint Agent to the various remote Webex servers used in that specific meeting, such as the multimedia platform and collaboration bridge.

Test Priority

An Endpoint Agent can run up to 10 concurrent scheduled tests and dynamic tests.
To ensure that the more relevant tests are executed when the total number of tests assigned to an agent exceeds 10, you can pre-define a priority for the tests.
In general, prioritization of tests is for dynamic tests; however, you can also prioritize scheduled tests.
Be sure to enable the Prioritize switch for all dynamic tests. For more information on how tests are assigned to agents, see Assign Tests to an Endpoint Agent.

Using the Endpoint Agent to Solve User Problems

This section describes how to use the Endpoint Agent to resolve user issues.
As a pre-requisite, ensure that the devices in your organization have the Endpoint Agent installed, a monitored domain set is active, scheduled tests are running, and dynamic tests are watching your critical applications.
There are three primary ways to start investigating endpoint issues:

Overview

The first place to investigate an issue is on the Endpoint Agents > Overview screen.
The Overview provides a high-level summary of both real user tests and scheduled tests. The default view shows you the worst results, allowing you to identify the issue quickly and start working on it.
All the URLs and test names in this view are clickable links that redirect you to the corresponding scheduled tests or real user tests. This enables you to follow up on generic issues in your environment more easily.

Agent Views

The helpdesk or IT support staff in your organization are the primary consumers of the Endpoint Agents > Agent Views screen.
In this view, you can search for a specific agent and see all the information related to that agent's scheduled tests, dynamic tests, local networks, and real user test data.
Using this view, you can easily click on the metric and jump to a detailed view that is pre-filtered for the selected agent.
For more information on Agent Views, see Endpoint Agent Views.

Detailed View

The Endpoint Agents > Views screen provides you with all the test details and results for all Endpoint Agents. Use the filter drop-down at the top of the screen to see the data associated with agents, destinations, networks, or VPNs. For more information on Views and how the different filters work, refer to the Viewing Data documentation.

Example of Troubleshooting

In this example, we walk through a real-life scenario of how to use the ThousandEyes Endpoint Agent for troubleshooting.
The user here experiences a sudden “slowness” when connecting to web applications, and as usual “they didn’t change anything”. This is a common complaint reported by all help desk personnel.
Use this snapshot to go through this example yourself.
Start your troubleshooting by selecting the Visited Pages view. In the image below, we see this screen, filtered by the agent belonging to the user with the complaints (lindsayc)
At 01:00, the moment shown, the experience looks good, matching what the user describes. The following supports that statement:
  • The average experience score is 95%, close to perfect.
  • The overall page speed is “Fast”.
  • There are no errors on the page.
The image below shows the same user one hour later. Here, we see the experience score is now 67%, significantly lower than the 95% earlier.
This information helps you to acknowledge the problem the user is experiencing. This is an important first step when your help desk is in conversation with an user who is experiencing application, network, or system-related issues.
The next step is identifying the root cause of the problem. Under Experience score by agent, click the agent name to zoom in on the details for the real user test in this specific time window.
The image above shows that the client is connected to the wireless network “AndroidAP” with a very low quality. The SSID name suggests that this is a cellular connection, and not office wifi, explaining the sudden drop in user experience.
You can find details about all the different view options for the Endpoint Agent in Viewing Endpoint Data.

Building Dynamic Dashboards Based on Labels

In Leveraging Endpoint Agent Labels, we used labels to determine which agent should run a specific test. Another, equally powerful use of labels is in dashboards. Use labels for grouping and selecting agents, resulting in dynamic dashboards.
In the image above, you can see the options to configure a new widget in a dashboard. As an example, we will compare wireless and wired application experience.
  1. 1.
    Log in to the ThousandEyes platform and go to Dashboards.
  2. 2.
    To create a new dashboard, click the three dots or ellipsis menu and select Create New Dashboard.
  3. 3.
    Once you've created your dashboard, click + Add Widget and select the Color Grid.
  4. 4.
    Fill in the fields as follows:
    • Widget Name: Wireless or Wired, per site
    • Data Source: Endpoint Real User Tests
    • Category: Visited Pages
    • Metric: Experience Score
    • Group Cards by: Endpoint Agent Label
    • Drill Down: Select the labels Wired and Wireless. (You must create the labels before creating this widget.)
  5. 5.
    Click Save.
The picture below shows the comparison between wired and wireless performance. In this case, wireless has a better performance across the entire organization.
For more information about ThousandEyes dashboards, see Dashboards.

Next Steps

After completing this getting-started guide, the following links will help you on the rest of your journey in getting the most out of the ThousandEyes Endpoint Agent.