~/talk

Tangram: Bridging Immutable and Mutable Abstractions for Distributed Data Analytics

on YouTube 21m
Talkat USENIX ATC '19 · on Wednesday, 2019-07-10· at 14:40· USENIX ATC Track II: Grand Ballroom VII–IX· [data-engineering, distributed-systems]

speakers

name:Yuzhen Huang
links:none on file
name:Xiao Yan
github:ImYanXiao
name:Guanxian Jiang
links:none on file
name:Tatiana Jin
links:none on file
name:James Cheng
links:none on file
name:An Xu
links:none on file
name:Zhanhao Liu
links:none on file
name:Shuo Tu
links:none on file

resources

video:https://youtu.be/o-tRDRZiHdc
slides:none on file

summary

Tangram implements MapUpdate to infer whether collections are mutable or immutable and lets update functions perform in-place, key-value updates while map functions remain functional. The system uses partition-based progress control and a bounded-delay synchronous model, plus context-aware failure recovery that distinguishes local vs global failures and either reloads immutable partitions or rolls back mutable partitions to the latest checkpoint. Implemented on 20 machines, Tangram matches specialized systems on bulk, graph, and iterative ML tasks and avoids costly pipeline checkpointing and cross-system data IO.

AI-written summary · about

chapters (9)

  1. 0:00Introduction
  2. 0:50Immutable vs Mutable
  3. 3:24Map Updates
  4. 8:07Map Updates Implementation
  5. 11:36Experiments
  6. 12:14Local failure recovery
  7. 12:55Global failure recovery
  8. 14:00Pipeline workload scenario
  9. 14:55Conclusion

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.