Like the title says I've been looking into this for some time now and haven't found any real solutions. I've tried out Sentry's profiling but it basically just showed overall memory usage which was nowhere near granular enough.

The main use case is when we have operations that use too much memory, I would like an easier way to identify what specifically is using that excess memory. Similarly, would like an easier way to identify the cause of memory leaks (even if its just pointing me in the right direction).

Any ideas would be appreciated. Thanks!