~/talk

Solving Max-Min Fair Resource Allocations Quickly on Large Graphs

on YouTube 17m
Talkat 21st USENIX Symposium on Networked Systems Design and Implementation (NSDI '24) · 2024-04-16..04-18· [distributed-systems]

speakers

name:Pooria Namyar
links:none on file
name:Behnaz Arzani
links:none on file
name:Srikanth Kandula
links:none on file
name:Santiago Segarra
links:none on file
name:Daniel Crankshaw
links:none on file
name:Umesh Krishnaswamy
links:none on file
name:Ramesh Govindan
links:none on file
name:Himanshu Raj
links:none on file

resources

video:https://youtu.be/0xiJEZbv-Rw
slides:none on file

summary

Converts the iterative max‑min fair solver into a single fast optimization by using epsilon weighting plus a sorting network, and introduces a multipath generalization of waterfilling plus a fast geometric binner that bins allocations geometrically and solves one weighted objective. The system offers an alpha‑approximation knob trading bounded unfairness for speed, empirically Pareto‑dominates prior methods, and is deployed in Azure WAN where it preserves solution quality while delivering roughly a 3× speedup.

AI-written summary · about

chapters (7)

  1. 0:00The resource allocation challenge
  2. 1:28Limitations of existing methods
  3. 3:15The Sur framework
  4. 5:30Technical approach
  5. 7:00Exact single-shot optimization
  6. 10:37Geometric binning
  7. 13:38Evaluation and conclusion

related talks

~/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.