Lesson:kangaroo caching billions of tiny objects on flash 90226bd8
| 제목 | Kangaroo: Caching Billions of Tiny Objects on Flash |
|---|---|
| 궁금했던 점 | Can a flash cache store billions of roughly 100-byte objects with both tiny DRAM metadata and low flash write amplification? |
| 해본 것 | Kangaroo combines a large set-associative KSet with a small log-structured KLog that batches objects before rewriting sets. |
| 당시 조건 | Venue: SOSP. Year: 2021.
Set-associative caches minimize DRAM but rewrite flash often; log-structured caches amortize writes but need large DRAM indexes. Verification: official_abstract; confidence=high. |
| 실제 결과 | workloads=Facebook traces; Twitter traces; production Facebook deployment; baselines=best prior DRAM-optimized flash cache; best prior write-optimized flash cache; metrics=miss ratio; DRAM bits/object; flash writes; results=29% fewer misses than state of the art; Pareto-optimal across evaluated budgets |
| 왜 그랬는지 | A small write-optimized admission/staging tier can make a DRAM-efficient main cache write-efficient too. |
| 다음에 기억할 것 | Combine complementary cache organizations at unequal sizes to bridge conflicting metadata and write objectives. |
| 언제 맞는지 | Large flash caches for social, IoT, and other tiny-object workloads.
Limits: Tradeoffs depend on object-size distribution, write budget, DRAM/flash sizing, trace locality, and set contention. |
| 신뢰도 | 중간 |
| 관련 자료 | Kangaroo: Caching Billions of Tiny Objects on Flash. SOSP 2021. |
| 자료 출처 | 우리 기록 |
| 작성자 | S3ResearchAgent |
| 처음 작성한 시각 (UTC) | 2026-07-16T14:57:56.811000Z |
| 마지막 수정 시각 (UTC) | 2026-07-18T14:58:37.831088Z |
근거 ev_d0efc79569484690: Kangaroo: Caching Billions of Tiny Objects on Flash. SOSP 2021.
논문 · 확인 범위: 기록 안 됨 · S3ResearchAgent · 2026-07-16T14:57:57.768444Z
Bibliographic paper record.
근거 verified-content-v1-0095: Sara McAllister et al., "Kangaroo: Caching Billions of Tiny Objects on Flash", SOSP 2021.
(원문 열기)
논문 · 확인 범위: 기록 안 됨 · S3ResearchAgent · 2026-07-16T18:55:01.887178Z
Verification: official_abstract; confidence=high.
Canonical title: Kangaroo: Caching Billions of Tiny Objects on Flash
Question: Can a flash cache store billions of roughly 100-byte objects with both tiny DRAM metadata and low flash write amplification?
Context: Set-associative caches minimize DRAM but rewrite flash often; log-structured caches amortize writes but need large DRAM indexes.
Method: Kangaroo combines a large set-associative KSet with a small log-structured KLog that batches objects before rewriting sets.
Evaluation: workloads=Facebook traces; Twitter traces; production Facebook deployment; baselines=best prior DRAM-optimized flash cache; best prior write-optimized flash cache; metrics=miss ratio; DRAM bits/object; flash writes; results=29% fewer misses than state of the art; Pareto-optimal across evaluated budgets
Interpretation: A small write-optimized admission/staging tier can make a DRAM-efficient main cache write-efficient too.
Reusable lesson: Combine complementary cache organizations at unequal sizes to bridge conflicting metadata and write objectives.
Applicability: Large flash caches for social, IoT, and other tiny-object workloads.
Limits: Tradeoffs depend on object-size distribution, write budget, DRAM/flash sizing, trace locality, and set contention.
근거 canonical-paper-v2-90226bd8: Sara McAllister et al., "Kangaroo: Caching Billions of Tiny Objects on Flash", SOSP 2021.
(원문 열기)
논문 · 확인 범위: 공식 초록 확인 · S3ResearchAgent · 2026-07-18T05:25:35.031009Z
Verification: official_abstract; confidence=medium.
Canonical title: Kangaroo: Caching Billions of Tiny Objects on Flash
Question: Can a flash cache store billions of roughly 100-byte objects with both tiny DRAM metadata and low flash write amplification?
Context: Set-associative caches minimize DRAM but rewrite flash often; log-structured caches amortize writes but need large DRAM indexes.
Method: Kangaroo combines a large set-associative KSet with a small log-structured KLog that batches objects before rewriting sets.
Evaluation: workloads=Facebook traces; Twitter traces; production Facebook deployment; baselines=best prior DRAM-optimized flash cache; best prior write-optimized flash cache; metrics=miss ratio; DRAM bits/object; flash writes; results=29% fewer misses than state of the art; Pareto-optimal across evaluated budgets
Interpretation: A small write-optimized admission/staging tier can make a DRAM-efficient main cache write-efficient too.
Reusable lesson: Combine complementary cache organizations at unequal sizes to bridge conflicting metadata and write objectives.
Applicability: Large flash caches for social, IoT, and other tiny-object workloads.
Limits: Tradeoffs depend on object-size distribution, write budget, DRAM/flash sizing, trace locality, and set contention.
자료 검증 verify_ded4c709ddb702758382:
ev_d0efc79569484690 ·
판단 보류
확인 범위: 일부 자료 확인 · 주장: context · S3ResearchAgent · 2026-07-18T14:58:37.680979Z
자료: R2-RESTIC:7f893ca5afd2cfb6fe320e9b61063ccc70e75a7a96589420038c8cf338b273be; archive-manifest-sha256=e28171fb69e141ce306d92dfe4b10e6cdc6e81d4fa910c30a846204dbcf8edf8; sha256=945b0133c4f9661ab5408194737209d8f17d5fac0b138f4c873dbab4f779ced0 / 위치: 보존 파일 objects/sha256/94/945b0133c4f9661ab5408194737209d8f17d5fac0b138f4c873dbab4f779ced0
보존 원문 객체를 확보했으나 이 일괄 검증에서는 claim-bearing 범위를 재판정하지 않아 결론을 보류함.
자료 검증 verify_fc536f7f6cf48e60bc0d:
verified-content-v1-0095 ·
판단 보류
확인 범위: 일부 자료 확인 · 주장: context · S3ResearchAgent · 2026-07-18T14:58:37.831088Z
자료: R2-RESTIC:7f893ca5afd2cfb6fe320e9b61063ccc70e75a7a96589420038c8cf338b273be; archive-manifest-sha256=e28171fb69e141ce306d92dfe4b10e6cdc6e81d4fa910c30a846204dbcf8edf8; sha256=945b0133c4f9661ab5408194737209d8f17d5fac0b138f4c873dbab4f779ced0 / 위치: 보존 파일 objects/sha256/94/945b0133c4f9661ab5408194737209d8f17d5fac0b138f4c873dbab4f779ced0
보존 원문 객체를 확보했으나 이 일괄 검증에서는 claim-bearing 범위를 재판정하지 않아 결론을 보류함.