When you decide to monitor your systems with Circonus, there’s quite a few options on how to collect your metrics. We believe Circonus should be a tool that does what you need, when you need it. Circonus does not force you into a specific approach or method. Since there are so many different ways to gather telemetry via Circonus, we thought we would take a moment to outline some of the different approaches.
In addition to application-specific checks, you might like to get baseline information about things like memory, CPU, file systems, and interfaces of your servers and network equipment. We’ve listed out the main options that can be used for system performance metrics below, along with a brief description and our recommendations for each. We’ve roughly ordered these based on our best practices, but the tool that should be used depends on many variables that you’ll need to take into account.
For instance, some users may prefer to use a single agent on all of their devices, which may mean that some options won’t be available. Available plugins and ability to expand should also be considered. Some agents allow Circonus to reach out to the endpoint and gather metrics, while others require the data to be pushed (these agents mention push requirements in the description below). In some cases, the language that the agent was written in can have an effect on your decision.
SNMP – SNMP is a standard that has been around for years, and allows monitoring of many types of network equipment, servers, and appliances. There is a good chance you already have SNMP configured on most of your hosts, which would significantly lower the up-front setup time. You’ll need to know the OIDs you want to monitor, but check bundle templates can make this process a little easier for you.
HTTPTrap – Circonus can accept JSON payloads via an HTTP PUT or POST request. This data is not polled regularly from the Circonus Broker, but is pushed to the Broker from the monitored target. This is the easiest way to get arbitrary data into Circonus, but you’ll have to figure out where to get the data.
collectd – Collectd is a lightweight C-based tool that has a variety of plugins available for data collection. There are 2 main ways to use collectd with Circonus, either to push the information from your device over UDP (similar to statsd and HTTP Traps) or via the write_http plugin.
Gollector – Gollector is a new monitoring agent that relies on the proc filesystem and C POSIX calls such as
sysconf to determine your machine’s profile. This alleviates any performance penalty from shelling out the collection work that some other agents can have.
NRPE – Circonus can utilize existing NRPE checks from your Nagios or Icinga installation. NRPE allows you to remotely call Nagios scripts to collect information. If you want to monitor a non-standard metric, there’s probably a Nagios script for it.
statsd – Similar to an HTTP Trap, statsd allows your hosts to send information to Circonus Enterprise Brokers, rather than the Broker reaching out to the host to poll it. One downside is that this information cannot be played in real-time, but it can be useful for metrics that may not have regular intervals of available information or are particularly high volume.
nad – Nad is a lightweight, simply managed host agent written in Node.js. Nad is the first choice of Circonus due to its easy extensibility and its ability to work on almost any platform, including Windows, RHEL, Ubuntu, and illumos derivatives. Nad comes with enough plugins to let you monitor any of the basics, while allowing you to add your own checks to fit your environment.
Resmon – Resmon is a Perl-based agent created by OmniTI. New modules can be created quickly and easily, but must be written in Perl; that’s a make it or break it factor for many.
Windows Agent – If you’d rather not use nad on your Windows servers, there is a Windows agent that can be used to collect performance metrics from Windows servers.
Which is right for me?
The choice of agent to use depends on many factors. Current operating system, existing monitoring setup, and network layout can all have an effect on which agent you choose. You may also need to incorporate several choices in order to best monitor your environment.
That covers the main ways to get system information into Circonus. There’s plenty of other methods of getting data, such as Google Analytics, a variety of database connections, Memcached, Varnish, NewRelic, and more. A combination of these collection types can enable you to have data on every piece of your infrastructure, so you can always find the information you need.