In today's video we'll explore how to create custom metrics with Blackfire, specially tailored to your application's unique needs.
Blackfire already provides over 600 built-in metrics. Sometime we need metrics specific to our own code and logic. That's when custom metrics become very useful.
Custom metrics are also stored in our .blackfire.yaml file. Suppose you want
to measure the time spent on a particular method like rendering your main
controller. You can create a custom metric for that.
This metric measures every calls to the MainController::render method. As
for any other metric, you will see its cardinality in the
Deterministic Profile UI,
which means how many times that metric has been identified, as well as its
contribution to all available dimensions.
In some cases you won't be interested in all that information. You can define
which contribution you are interested in with the contrib key. Are you
interested in the count only? The cost only? The cost being the contribution to
all dimensions. Or even both of them, which is the default option.
Timeline metrics are those metrics listed at the top of the metric list and
associated with a colored background to be better identified. You can have your
custom metrics in that list by adding the timeline: true option.
Another option to improve the visibility of a custom metric is to have a marker
displayed every time the metric is detected. This is possible by adding the
marker option with the description of that marker. This is a great
debugging tool.
Our custom metrics can be precise and super flexible at the same time. Not only can we define one or more callees, which are the functions being called, but you can define callers as well.
You can target when a specific function is being called by another specific method. This allows you laser-focused performance insights.
You can go even beyond this by using regular expressions. You could even target classes implementing a specific interface. This is truly a powerful feature and the Jedi way of mastering Blackfire. Check the documentation under Tests > Metrics to discover all the options available.
In that documentation, you will learn, for instance, about argument capturing, giving you the possibility to group function calls by specific arguments. This is a great way to bring more clarity to your profile information.
Finally, if your customer metrics start growing, a good practice is to organize them with layers. What about grouping the custom metrics for your databases operation, user management, or billing?
The layer key is your option to create, well, layers. Note that those layers
can be nested as well. The billing layer can be split into
billing.invoice, billing.refund, themselves could also hold more nested
layers.
Using custom metrics allows you to measure exactly what matters to your application. They make your performance test more reliable and help you enforce higher standards.
Your next step is to create your first custom metric. Then we will see how to start automating performance testing to secure your critical customer journeys.