The system architecture required to install SaltStack Config depends on two main factors: 1) the installation method you are using to deploy SaltStack Config and 2) your environment's throughput, meaning the amount of work that you will perform on your system using SaltStack Config.
Before you start
In order to make an accurate assessment of your system architecture needs, you should first ensure you are familiar with:
- The two available installation methods for SaltStack Config
- The four basic components of the SaltStack Config SaltStack architecture (RaaS, the Salt master, PostgreSQL, and Redis)
See Installing and Configuring SaltStack Config for an overview of these concepts, including a general overview of the installation process. Also see Which installation scenario should you use? for guidance selecting an installation scenario.
SaltStack Config is powered by Salt, an open-source automation and configuration management engine. If you are less familiar with key terms used in Salt (such Salt master and Salt minion), see Salt system architecture for more information.
Determining minion architecture
In the context of SaltStack Config, a minion generally refers to a node in your production environment that connects with and is managed by SaltStack Config through one or more Salt masters.
Salt is designed to work with any operating system that might be running on a minion. In addition to the standard operating systems (Linux, Windows, MacOS), Salt provides specialized minion software (generally referred to as "native minions") for operating systems that are unique to various network devices such as Arista, Juniper, AIX, and Solaris.
This table lists the minimum memory requirements for the Salt minion service by operating systems:
|Operating system||Minimum memory requirements|
|AIX minion||512 MB RAM|
|MacOS minion||4 GB RAM|
|Linux minion||512 MB RAM|
|Windows minion||4 GB RAM|
|Other network devices, including proxy minions||40 MB RAM per controlled device|
Determine your installation scenario
See Which installation scenario should you use? for guidance selecting an installation scenario. If you are unsure which installation method is best for your system, the standard installation is recommended. The Lifecycle Manager installation method is not recommended for production grade systems with more than 1,000 nodes.
If you choose a Lifecycle Manager installation, it requires only one node and needs the following system architecture:
|Hardware||Up to 1,000 nodes (minions)|
|Cores||16 CPU cores|
|RAM||32 GB RAM|
|Disk space||At least 260 GB free space|
The remainder of this guide will explain architecture needs for the standard installation scenario.
Estimate the number of Salt minions you will manage
Although the throughput of your system is difficult to measure prior to installation, you can estimate your needs based on the number of minions (nodes) in your system that will be managed by SaltStack Config. The last section of this guide provides additional measurements for determining your system throughput.
As you bring more Salt minions under management by SaltStack Config, you might need to increase your system architecture to match.
Estimate the number of Salt masters you need
This table lists the recommended number of Salt masters you might need based on the number of managed Salt minions (nodes) in your system:
|Minions||Salt masters (16 CPU/16 GB)|
This document describes SaltStack Config on-premises architecture. SaltStack Config Cloud can currently only run one Salt master, which means it is limited to less than 20,000 minions.
Estimate the number of RaaS nodes you need
This table lists the recommended number of RaaS nodes you might need based on the number of managed Salt minions (nodes) in your system:
|Minions||RaaS nodes with 16 CPU/16 GB||RaaS nodes with 32 CPU/32 GB|
Estimate the number of PostgreSQL nodes you need
The next two tables list the recommended number of PostgreSQL database nodes you might need based on the number of managed Salt minions (nodes) in your system:
|Minions||PostgreSQL nodes with 8 CPU/8 GB||PostgreSQL nodes with 16 CPU/16 GB||PostgreSQL nodes with 24 CPU/24 GB||PostgreSQL nodes with 32 CPU/32 GB|
|Minions||PostgreSQL nodes with 48 CPU/48 GB||PostgreSQL nodes with 56 CPU/56 GB||PostgreSQL nodes with 64 CPU/64 GB|
Estimate the number of Redis nodes you need
The next two tables list the recommended number of Redis database nodes you might need based on the number of managed Salt minions (nodes) in your system:
|Minions||Redis nodes with 4 CPU/4 GB||Redis nodes with 8 CPU/8 GB||Redis nodes with 12 CPU/12 GB|
|Minions||Redis nodes with 16 CPU/16 GB||Redis nodes with 20 CPU/20 GB|
Optimize your architecture after installation based on throughput
After you've completed your SaltStack Config installation, you can use system monitoring metrics to better determine your system's throughput and architectural needs.
When determining what to monitor, consider these factors:
- Amount of traffic on the event system - The event system (also sometimes referred to as the "event bus") is used for inter-process communication by both Salt master and Salt minions. If your event bus is very busy, consider inceasing your memory allocations.
- Job returns per hour - SaltStack Config uses the term "jobs" to refer to each of the commands, tasks, and operations performed by SaltStack Config. Each job sends its output to SaltStack Config for reporting and data collection purposes. The number of job returns your system produces in a given hour can impact your architecture needs.
- Amount of pillar data - Pillar data is data that must be stored on the Salt master. Pillar is primarily used to store secrets or other highly sensitive data, such as account credentials, cryptographic keys, or passwords. Pillar is also useful for storing non-secret data that you don't want to place directly in your state files, such as configuration data. The amount of data being stored on your Salt master (and later accessed by minions as needed) can impact your memory and data storage needs.
- Amount of custom grains - Grains are used in Salt to target the minions for a particular job or command. Grains refer to the basic data and characteristics of each minion. Salt comes with many pre-built grains. For example, you can target minions by their operating system, domain name, IP address, kernel, memory, and many other system properties. You can also create custom grain data to distinguish one group of minions from another based on a characteristic you uniquely target for in your system. The number of custom grains you create can impact your architecture needs.
- Number of beacons and reactors - The beacon system is a monitoring tool that can listen for a variety of system processes on Salt minions. Beacons can trigger reactors, which can then help implement a change or troubleshoot an issue. For example, if a service’s response times out, the reactor system can restart the service. When coupled with reactors, beacons can create automated, pre-written responses to infrastructure and application issues. Reactors expand Salt with automated responses using pre-written remediation states. If your system has beacons and reactors that are activated regularly, that could increase your system architecture needs.
- Disk size needs - You might need to increase your disk size based on the number of minions being managed and for each year's worth of data you need to keep in storage. For example, if you have a high-throughput system and your system is in a highly-regulated industry that requires 7-8 years of data retention, that might require higher disk size and storage capacity.
- Geographical location and distance between components - You might experience issues if there is a 65 ms latency or more between the Salt master and the server that is running SaltStack Config (RaaS). Fortunately, Salt is less sensitive to latency between the Salt minion and Salt master. When placing these components, keep in mind that it is better to locate the master close to RaaS and minion farther away if necessary.
- Business-critical operations - When assessing how business-critical SaltStack Config is in your environment, ask yourself how severe the impact of a SaltStack Config outage would be to your business. If it were down for an hour or more, would it cause a severe impact? If so, you might need to design high availability needs into your SaltStack Config system architecture.
Based on these factors, consider incrementally increasing your resources and monitor the impact to your system's performance. For example, increasing your memory allocations by 4GB RAM with 4 CPUs.
The following image shows an example of a high availability SaltStack Config architecture design:
As this image illustrates, many high availability systems connect to multiple Salt masters. High availablility systems also often build redundancy into the PostgreSQL database and Redis databases so that one can fail over to another. Keep in mind that the current high availability solutions for PostgreSQL and Redis only support manual failovers.