Installation Strategies
Sensu’s architecture is one of its most compelling features. It is flexible enough to be installed on a single system for development/testing/lab purposes (or small production environments), and sophisticated enough to support highly available configurations capable of monitoring infrastructure at scale.
Please review the following definitions of standalone, distributed, and high-availability installation strategies to help you select which one will be the most appropriate for your installation. If you’re just getting started with Sensu and/or if you’re not sure which strategy to choose, follow the instructions for a standalone installation.
Standalone
Install all of Sensu’s dependencies and services on a single system. For the purposes of this installation guide (which is designed to help new users learn how Sensu works and/or setup Sensu in a development environment), a standalone installation is recommended.
To proceed with a standalone installation, please select a single compute resource with a minimum of 2GB of memory (4GB recommended) (e.g. a physical computer, virtual machine, or container) as your installation target, and continue to the next step in the guide.
NOTE: Sensu’s modular design makes it easy to upgrade from a standalone installation to a distributed or high-availability installation, so unless you have some specific technical requirement that demands a distributed or high availability installation, there’s usually no need to start with a more complex installation.
Distributed
Install Sensu’s dependencies (e.g. RabbitMQ and/or Redis) and services (i.e.
the Sensu server and API) on separate systems. The only difference between
a Standalone installation and a Distributed installation is that Sensu’s
dependencies and services are running on different systems. As a result,
although this guide will explain how to perform a Distributed Sensu
installation, it will not cover such industry-standard concepts as networking,
etc (i.e. configuring services to communicate with other services installed
elsewhere on the network will be left as an exercise for the user; e.g.
replacing default localhost configurations with the corresponding addresses
and/or ports, and ensuring that the appropriate network connections and firewall
rules will allow said services to communicate with one another).
To proceed with a distributed installation, please select a minimum of two (2) compute resources (e.g. physical computers, virtual machines, or containers) as your installation targets, and continue to the next step in the guide.
NOTE: for the purposes of this installation guide, distributed installation will be described in terms of two (2) installation targets. One system will act as the “transport and datastore” system, and one system will be act as the Sensu server. Advanced users who may wish to use more than two systems are welcome to do so (e.g. using one as the transport/RabbitMQ, one as the data store/Redis, one as the Sensu server, and one or more for running Sensu clients).
High Availability
Install Sensu’s dependencies across multiple systems, in a high-availability configuration (clustering, etc), and install the Sensu services on multiple systems in a clustered configuration. High availability configurations will be introduced at conclusion of this guide.
