Was this page helpful?
Yes No

Performance Tuning Examples

Tableau Server on Windows now includes Tableau Services Manager (TSM), which replaces the Configuration Utility and the command line tool. If you need help for an earlier version of Tableau Server, see the Tableau Help page.

This topic lists example process configurations for Tableau Server installations with one, two, and three nodes. Use these process configurations as a starting point when tuning the number of server processes in your installation of Tableau Server.

In this topic:

One-node example: Balanced

This example shows a 64-bit, 8+ core, 16+ GB system configured for heavy extract usage.

The configuration for this example would look like this:

Configuration notes

  • The server runs two VizQL Server processes, two Cache Server processes, and two Data Server processes. These are the recommended values and are the defaults from installation.

  • As a general rule, run a Cache Server process for every VizQL Server process on the node.

  • Calculate the minimum number of Backgrounder processes to run by dividing the computer's total number of physical cores by 4. To calculate the maximum number, divide the computer's total physical cores by 2.

  • Both the Backgrounder and Data Engine processes are CPU-intensive.

  • Schedule extract refreshes for off-peak times to help the VizQL Server, Application Server, Data Engine, and Backgrounder processes to not compete for system resources.

Two-node example: Optimized for heavy extract usage

This example shows a possible configuration for a two-node Tableau Server deployment that handles heavy extract usage. Heavy extract meaning many extracts and frequent extract refreshes. Your backgrounder are busy all the time, during peak hours and non-peak hours.

Both the nodes are on 64-bit, 8+ core, 16+ GB systems.

Note that the VizQL Server, Application Server, and Data Server processes on the initial node are isolated from the background processes, which are running on the additional node.

The configuration for this example would look like this:

Configuration notes

The main goal here is to isolate backgrounder process on its own node to facilitate extract refreshes. Backgrounder and the co-located Data Engine are used during refresh extracts, so isolating them on a separate node will help improve remove contention between extract refresh process and user views and interactions with Workbooks.

  • The initial node runs two VizQL Server processes, two Cache Server processes, and two Data Server processes.

  • As a general rule, run a Cache Server process for every VizQL Server process on the node.

  • Isolate the Backgrounder processes by configuring them to run on the additional node. To calculate the minimum number of Backgrounder processes to run, divide the computer’s total number of physical cores by 4. To calculate the maximum number, divide the computer's total physical cores by 2.

  • Isolate the Backgrounder processes from the VizQL Server, Application Server, and Data Server processes.

  • Separate the File Store process from the backgrounder to help minimize the backgrounder processes from affecting user views.

Three-node example: Optimized for a balance between extracts and user traffic

A configuration of three nodes or more is recommended to achieve the best performance when you have a high amount of extract refreshing and usage, and a high number of concurrent users. In this example, all computers are assumed to be 64-bit, 16 core, 16+ GB systems.

The configuration for this example would look like this:

Configuration Notes

  • For this configuration, 16 cores are recommended for each node.

  • Run two VizQL Server processes, two Cache Server processes, and two Data Server processes on the nodes that are not running Backgrounder processes.

  • As a general rule, run a Cache Server process for every VizQL Server process.

  • The Backgrounder processes are on their own node so that they do not compete for resources with the other processes. Because this node is dedicated to Backgrounder processes and they might consume 100% of the CPU resources, the recommended number of Backgrounder processes is the number of physical cores divided by 2. However, for a more extract heavy environment, you might want to increase the number of Backgrounder processes to equal the number of physical cores for the node.

  • Run File Store processes on the initial node and on the additional node (shown as Worker 1 in the Server Status image above) that is not running Backgrounder processes. This allows Tableau Server to split view requests between the two nodes.

  • The user loads for the Application Server and Data Server processes can typically be handled by a single instance of each process. However, you can configure two of each process to provide redundancy.

  • Under most conditions, the Data Server processes and the Data Engine processes will not be a bottleneck for the system’s overall throughput as long as sufficient CPU cycles exist for them. To increase viewing capacity, add additional nodes and run dedicated VizQL Server processes on them. To increase capacity for refreshing extracts, add additional nodes and run dedicated Backgrounder processes on them.