A call-graph is a relational visualization method. Blackfire profiles are
displayed as call-graphs in order to show the caller-callee relation between
the function/method calls, which shows best the code’s behavior. It enables for
instance to understand whether a specific function call is responsible for a
bottleneck, or if the bottleneck comes from their parents.
A callee is a function/method which is called by another function/method in
A caller is a function/method which calls another function/method in your code.
The CPU Time is a profiling dimension in Blackfire. It corresponds to the time
your server’s CPU actually spends executing your code.
The exclusive time is the time spent to execute a function, excluding the
time spent executing its callees. The exclusive time tells you which
functions consumed the most time by themselves; those are probably the ones
you might want to optimize first.
The I/O time corresponds to the time that the CPU is waiting for input/output
operations. That can for instance correspond to network (database calls, Redis
calls, HTTP calls, …) or disk activity (file inclusions, …). Keep in mind
that in Blackfire profiles the I/O is almost never 0 as it also includes
non-significant activities like memory access.
The inclusive time is the time spent to execute a function, including all of
A node is a call-graph element. In Blackfire, it represents a function/method
call, or an aggregation of calls.
An Organization enables to have multiple projects with different
subscription levels. They give more flexibility for companies working on
several projects in parallel to tune the amount of users and
environments for each project, the
subscription level, and the billing cycle.
A profile is the data output while measuring the resource consumption of an
executed piece of code. In Blackfire, that includes consumption in Wall-time,
CPU time, I/O time, Memory, Network, HTTP calls and SQL queries.
Part of the data preprocessing done by Blackfire involves removing
function calls that take less that 1% of the global costs on all
dimensions (time, cpu, memory, etc.). This is done to reduce the size of the
profiles and to provide a responsive UI. You won’t be able to find these removed
nodes in the call graph nor in the functions list. Pruning can be disabled for
The wall time for a function call is the measure of the real time it took to
execute the code. It’s the difference between the time at which the function
was entered and the time at which the function was left. Its is actually the
sum of CPU time + I/O time.