Pyroscope identifies performance issues in your application by continuously profiling the code.
If you've never used a profiler before, then welcome!
If you are familiar with profiling and flame graphs, then you'll be happy to know that Pyroscope:
- Has low CPU overhead thanks to sampling profiler technology
- Can store years of profiling data down to 10 second granularity
- Efficient compression, low disk space requirements
There are two main components that allow Pyroscope to run smoothly and quickly:
Pyroscope agent wraps around your Python, Ruby, or Go application and polls the stacktrace 100 times per second to calculate which functions are consuming CPU.
Pyroscope Agent records and aggregates what your application has been doing, then sends that data to the Pyroscope Server over port
:4040 to be processed, aggregated, and stored for speedy queries of any time range. Later you can look at the profiling data and make queries across any time range for example:
- all of 2020
- that one day last month when the website was slow
- that time you deployed on a Friday night and messed up everything without knowing why
Check out our Demo Page and select any time range to see how quickly Pyroscope works!