Exploiting Combined Locality for Wide-Stripe Erasure Coding in Distributed Storage
summary
Proposes combined locality that mixes parity locality and topology locality, constructs stripes by placing HRC-coded chunks across racks with c=f, and implements Easy-Wide with multi-node encoding and indirect parity updates. Builds two prototypes (cold and edge) and runs experiments on Amazon EC2 showing single-chunk repair time drops up to 90.5% versus prior locality schemes while keeping redundancy as low as 1.063x; also reports higher full-node repair throughput and heuristics to avoid requester overload during parallel repairs.
AI-written summary · about
chapters (13)
- 0:00Intro
- 1:05Wide-stripe Erasure Coding
- 1:48Challenges for Wide Stripes
- 2:43Locality in Erasure-coded Repair
- 3:47Existing locality schemes for wide str
- 4:21Motivating Example
- 4:57Our Contributions
- 5:34Combined Locality
- 6:59Trade-off Analysis
- 8:01Repair in ECWide
- 9:06Implementation
- 9:25ECWide-H Experiments
- 10:02Conclusions
related talks
15mLESS is More for I/O-Efficient Repairs in Erasure-Coded Storage
11mStripeless Data Placement for Erasure-Coded In-Memory Storage
16mBoosting Full-Node Repair in Erasure-Coded Storage
17mParaRC: Embracing Sub-Packetization for Repair Parallelization in MSR-Coded Storage
51mConvertible codes: Adaptive coding for large-scale data storage
15mSelectiveEC: Selective Reconstruction in Erasure-coded Storage Systems
01h 14mErasure Coding at Scale
59mPractical Error Control Coding For Data Storage