The ThousandEyes platform uses agents to run tests against targets configured for measurement. An agent is a Linux server running custom ThousandEyes software, which checks in with an agent collector to obtain instructions from the ThousandEyes platform. Generally speaking, these are lightweight machines that are tasked solely with acting as ThousandEyes agents.
For a discussion of the different agent types, see Comparison of Agent Types.
An Enterprise Agent is an endpoint that is used to test targets from inside your network, or from infrastructure within your control. Enterprise Agents can be easily installed on your own network, in data centers and branch offices, using a lightweight virtualized system called a virtual appliance, a Docker container, or a Linux package installed on a supported Linux distribution.
Enterprise Agents collect data for the exclusive use of your account, and are not shared with other organizations.
The virtual appliance is a lightweight all-in-one package, intended for deployment into a hypervisor platform. The virtual appliance contains a web-based management console, which allows you to configure network, proxy, and other settings specific to this virtual appliance. One of the biggest benefits of the virtual appliance is that it is manageable by someone with little to no Linux expertise.
To install a virtual appliance, you must have a hypervisor available. Check out this article for suggestions on enabling an appropriate hypervisor.
The virtual appliance runs Ubuntu Linux server, and keeps itself up to date by downloading and applying security updates automatically. It is available in two different distribution options:
The industry-standard .ova template format
A Microsoft Hyper-V .zip file
Both distribution types can be downloaded from the Add > New Agent page.
In addition, you can preconfigure virtual appliances for your exclusive use. This binds your account token to the agent, such that it cannot be modified - and requires less configuration during Enterprise Agent deployment.
The Linux-package Enterprise Agent runs on any compatible version of Linux. For the current list of Linux distributions that support the Enterprise Agent Linux package, check Supported Agent Operating Systems.
Once the package is installed on the host and configured, it runs as a service. For Enterprise Agents that are installed using the Linux-package approach, the only software that is kept up to date is the software required by the agent code.
For instructions on installing the agent using a Linux package, see Linux Package Method.
Enterprise Agents can be grouped into agent clusters. Clusters allow you to add capacity to a single logical agent as needed.
When a test is assigned to the cluster, the cluster software assigns the test to the Enterprise Agent in the cluster that has the least load for the type of test being assigned. Note that clustering does not provide a failover mechanism. If one member of the cluster goes offline or experiences similar issues, the tests of that cluster member are not reallocated to the other members of the cluster.
Cluster members can be either virtual appliances or Linux-package Enterprise Agents.
For more information on agent clusters, see Working with Enterprise Agent Clusters.
For details, see Enterprise Agent Hardware Requirements.