~/talk

HeapBuffers: Why not just using a binary serialization format for your managed memory?

Talkat OOPSLA 2025 · 2025-10-16..10-18· [web-performance]
links:none on file
links:none on file
links:none on file
links:none on file
slides:none on file

HeapBuffers stores Java objects off-heap FlatBuffers layout in a separate transferable heap, generates schema-specific GC and specialized Java code, and compacts/encodes on prepare so the memory image is immediately IO-ready. Cross-heap references are disallowed and strings are copied via the Java heap, which can be the main overhead. Benchmarks against Apache Fury show large wins—often orders of magnitude for IO and in one set 2.4x throughput / 4x latency—with a worst case of about 8% throughput loss when schemas are string-heavy.

AI-written summary · about

  1. 0:00HeapBuffers intro
  2. 2:26Flat buffers in JVM
  3. 3:51HeapBuffers design
  4. 6:41HeapBuffers programming
  5. 9:53HeapBuffers performance
  6. 11:41Conclusion
  7. 12:23Q&A
~/search
type to search 6,249 conferences, 138,987 speakers and 185,172 talks
~/help
cmdctrl k
search every conference, speaker and series
/
focus the filter search
h
home
c
conferences
s
speakers
e
series
t
light / dark — the status bar button does the same
?
this help
esc
close this, then clear every filter — or just the search, while you are typing in it

Ignored while you are typing. Pages without a filter rail have no search to focus.