StackState uses Graphite data to create a total unified overview of the IT stack. On this, it supplies functions like investigation, remediation and prevention by using big data analytics on top of its real-time operations model. This will give you faster mean times to repair. It also gives you the ability to know the effect of changes even before they are applied.
The benefits of using Graphite data in StackState
StackState supports streaming monitor data of two kinds: event data and metric data. These streams are normalized. Thus any combination of streaming data, no matter where they originate from, can be used for investigation. A stream or combination of streams can be shown. They can be used for alerting, and followed back to their origins. StackState can create a live metric stream out of any Graphite query. That means that any Graphite metric data, including those that are transformed on the fly by Graphite, get all the benefits that metric streams have in StackState.
After telling StackState which Graphite endpoints are available arbitrary metric streams can be bound to any component or relation in the stack. In the user-interface this is achieved by simply dragging a Graphite metric on top of a component or relation. Anything that can be achieved via the user-interface can also be automated via the REST api, the synchronization wizards or our import tool.
So using Graphite data in StackState:
- Adds great alerting capabilities to Graphite
- Enables easy comparison of multiple Graphite metrics and events
- Correlates time series metrics with the stack
How to use Graphite data in StackState
Let’s see what we can do with a metric from Graphite in StackState. We will model a very simple example stack that hosts a website. The website’s traffic is routed via a load-balancer, which balances between three web-servers that share a replicated database.
In this example we will bind a Graphite metric that represents network throughput gathered by collectd to a relation in StackState which models the network connection between a webserver and a load-balancer. Clicking on the relation in StackState’s stack visualization will immediately give you a live preview of the throughput during the last hour. The graph can be further inspected using StackState’s built-in metric inspector.
Besides being able to see the metrics by visually navigating to the element in the stack, metrics can also be used as an input for running checks. Checks are functions, which are pre-supplied. But they may also be user-defined, that receive one or more monitoring streams as input and produce states as output. Using the “period below threshold” function, we can create a check for our network throughput metric. Thus if throughput ever falls to zero over more than half a minute, the network connection will report an error. The load-balancer and website, which are dependent on the connection are then automatically alerted, because of StackState’s automatic error propagation model.
In this case there is a cluster of applications running and just one link does not work. The propagation setting for the load balancer is set to propagate error states only if more than half the backend servers are in a failure state. But a warning is reported. This makes sure automatic actions can be taken and/ or a notification can be send.
We have made it extremely easy to connect your Graphite metrics streams to StackState. With the power of our IT operations platform you’re able to integrate with multiple monitoring solutions. This gives you the ability to run checks over any kind of monitoring data in any combination, taking the entire stack into account. Do you also want to hook up your Graphite data to StackState? Sign up for the download.