11th Symposium on Software Performance 2020 (SSP´20)
Memory anomalies such as memory leaks can dramatically
impact application performance and can even
lead to crashes. Thus, supporting developers in understanding
the heap memory behavior of their systems is
essential. Unfortunately, most memory analysis tools
lack advanced visualizations that could facilitate developers
in analyzing suspicious memory behavior.
To analyze heap memory, it is common to group
the heap's objects, for example, by their types or by
their allocation sites. Using multiple grouping criteria
thus results in a tree-shaped representation of the
heap content. Such a heap tree is then typically presented
textually in a tree table.
In this paper, we present ongoing research on using
well-known tree visualization techniques to visualize
such heap trees as well as their evolution over time.
Such visualizations may ease the detection of proliferating
heap objects, facilitating memory leak analysis.
To demonstrate the feasibility and applicability of
the presented approach, we implemented a web-based
visualization tool and integrated it into AntTracks,
our trace-based memory monitoring tool.