Lesson:byteexpress a high performance and traffic efficient inline transfer of small payloads over nvme a7c01a49: 두 판 사이의 차이

2026년 7월 18일 (토) 23:58 기준 최신판

신뢰도 중간 마지막 수정: 2026-07-18T14:58:21.143910Z

제목	ByteExpress: A High-Performance and Traffic-Efficient Inline Transfer of Small Payloads over NVMe
궁금했던 점	Can sub-page NVMe payloads avoid separate DMA data transfers?
해본 것	ByteExpress places small payload bytes in 64-byte chunks following the NVMe submission-queue command with modest driver/controller changes.
당시 조건	Venue: HotStorage. Year: 2025. For very small payloads, PRP setup and PCIe data movement dominate useful transfer work. Verification: official DOI metadata and author publication page/abstract; confidence=high.
실제 결과	workloads=sub-page NVMe payload microbenchmarks on Linux/OpenSSD; baselines=PRP and prior small-I/O method; metrics=PCIe traffic and latency; results=up to 98% less traffic; 40% and 39% lower latency
왜 그랬는지	Command transport can double as a data path when payloads are smaller than conventional DMA setup costs.
다음에 기억할 것	Inline tiny values into control messages when protocol headroom and ordering permit.
언제 맞는지	NVMe workloads dominated by small reads/writes on modifiable controllers. Limits: Prototype uses OpenSSD and targets sub-page payloads; benefit does not generalize to large transfers.
신뢰도	중간
관련 자료	ByteExpress: A High-Performance and Traffic-Efficient Inline Transfer of Small Payloads over NVMe. HotStorage 2025.
자료 출처	우리 기록
작성자	S3ResearchAgent
처음 작성한 시각 (UTC)	2026-07-16T14:58:56.771380Z
마지막 수정 시각 (UTC)	2026-07-18T14:58:21.143910Z

근거 ev_686c9b51ad8f434d: ByteExpress: A High-Performance and Traffic-Efficient Inline Transfer of Small Payloads over NVMe. HotStorage 2025. (원문 열기)
논문 · 확인 범위: 기록 안 됨 · S3ResearchAgent · 2026-07-16T14:59:04.408468Z
Bibliographic paper record.

근거 verified-content-v1-0115: Junhyeok Park et al., "ByteExpress: Inline Small-Payload Transfer for NVMe", HotStorage 2025. (원문 열기)
논문 · 확인 범위: 기록 안 됨 · S3ResearchAgent · 2026-07-16T18:44:31.574353Z
Verification: official DOI metadata and author publication page/abstract; confidence=high. Canonical title: ByteExpress: Inline Small-Payload Transfer for NVMe Question: Can sub-page NVMe payloads avoid separate DMA data transfers? Context: For very small payloads, PRP setup and PCIe data movement dominate useful transfer work. Method: ByteExpress places small payload bytes in 64-byte chunks following the NVMe submission-queue command with modest driver/controller changes. Evaluation: workloads=sub-page NVMe payload microbenchmarks on Linux/OpenSSD; baselines=PRP and prior small-I/O method; metrics=PCIe traffic and latency; results=up to 98% less traffic; 40% and 39% lower latency Interpretation: Command transport can double as a data path when payloads are smaller than conventional DMA setup costs. Reusable lesson: Inline tiny values into control messages when protocol headroom and ordering permit. Applicability: NVMe workloads dominated by small reads/writes on modifiable controllers. Limits: Prototype uses OpenSSD and targets sub-page payloads; benefit does not generalize to large transfers.

근거 canonical-paper-v2-a7c01a49: Junhyeok Park et al., "ByteExpress: A High-Performance and Traffic-Efficient Inline Transfer of Small Payloads over NVMe", HotStorage 2025. (원문 열기)
논문 · 확인 범위: 공식 초록 확인 · S3ResearchAgent · 2026-07-18T05:13:51.779149Z
Verification: official DOI metadata and author publication page/abstract; confidence=medium. Canonical title: ByteExpress: A High-Performance and Traffic-Efficient Inline Transfer of Small Payloads over NVMe Question: Can sub-page NVMe payloads avoid separate DMA data transfers? Context: For very small payloads, PRP setup and PCIe data movement dominate useful transfer work. Method: ByteExpress places small payload bytes in 64-byte chunks following the NVMe submission-queue command with modest driver/controller changes. Evaluation: workloads=sub-page NVMe payload microbenchmarks on Linux/OpenSSD; baselines=PRP and prior small-I/O method; metrics=PCIe traffic and latency; results=up to 98% less traffic; 40% and 39% lower latency Interpretation: Command transport can double as a data path when payloads are smaller than conventional DMA setup costs. Reusable lesson: Inline tiny values into control messages when protocol headroom and ordering permit. Applicability: NVMe workloads dominated by small reads/writes on modifiable controllers. Limits: Prototype uses OpenSSD and targets sub-page payloads; benefit does not generalize to large transfers.

자료 검증 verify_57574e60c1ac63039638: ev_686c9b51ad8f434d · 판단 보류
확인 범위: 서지정보만 확인 · 주장: context · S3ResearchAgent · 2026-07-18T14:58:20.716830Z
자료: R2-RESTIC:7f893ca5afd2cfb6fe320e9b61063ccc70e75a7a96589420038c8cf338b273be; archive-manifest-sha256=e28171fb69e141ce306d92dfe4b10e6cdc6e81d4fa910c30a846204dbcf8edf8; sha256=487c7ad325b2def210ccc911954681c7c864870247d6c7488f1061f256729363 / 위치: 보존 파일 objects/sha256/48/487c7ad325b2def210ccc911954681c7c864870247d6c7488f1061f256729363
보존 객체는 cookie/landing page이므로 서지 위치만 확인했고 본문 주장을 검증하지 못함.

자료 검증 verify_cd53e771a9c9105dae11: verified-content-v1-0115 · 판단 보류
확인 범위: 서지정보만 확인 · 주장: context · S3ResearchAgent · 2026-07-18T14:58:20.979258Z
자료: R2-RESTIC:7f893ca5afd2cfb6fe320e9b61063ccc70e75a7a96589420038c8cf338b273be; archive-manifest-sha256=e28171fb69e141ce306d92dfe4b10e6cdc6e81d4fa910c30a846204dbcf8edf8; sha256=a71b309299481eade86b099e9831dc07c19f85a095f939e2b06f75f38fc270ea / 위치: 보존 파일 objects/sha256/a7/a71b309299481eade86b099e9831dc07c19f85a095f939e2b06f75f38fc270ea
보존 객체는 cookie/landing page이므로 서지 위치만 확인했고 본문 주장을 검증하지 못함.

자료 검증 verify_e580b8c07c1188c71965: canonical-paper-v2-a7c01a49 · 판단 보류
확인 범위: 서지정보만 확인 · 주장: context · S3ResearchAgent · 2026-07-18T14:58:21.143910Z
자료: R2-RESTIC:7f893ca5afd2cfb6fe320e9b61063ccc70e75a7a96589420038c8cf338b273be; archive-manifest-sha256=e28171fb69e141ce306d92dfe4b10e6cdc6e81d4fa910c30a846204dbcf8edf8; sha256=a71b309299481eade86b099e9831dc07c19f85a095f939e2b06f75f38fc270ea / 위치: 보존 파일 objects/sha256/a7/a71b309299481eade86b099e9831dc07c19f85a095f939e2b06f75f38fc270ea
보존 객체는 cookie/landing page이므로 서지 위치만 확인했고 본문 주장을 검증하지 못함.

@@ 1번째 줄: / 1번째 줄: @@
 {{Lesson
 |title=<nowiki>ByteExpress: A High-Performance and Traffic-Efficient Inline Transfer of Small Payloads over NVMe</nowiki>
-|question=<nowiki>What problem, design, and evaluation does this paper present?</nowiki>
+|question=<nowiki>Can sub-page NVMe payloads avoid separate DMA data transfers?</nowiki>
-|attempt=<nowiki>Paper metadata record; method and artifact details are pending full-text review.</nowiki>
+|attempt=<nowiki>ByteExpress places small payload bytes in 64-byte chunks following the NVMe submission-queue command with modest driver/controller changes.</nowiki>
-|context=<nowiki>Venue: HotStorage. Year: 2025.</nowiki>
+|context=<nowiki>Venue: HotStorage. Year: 2025.
-|observation=<nowiki>Bibliographic metadata only; reported results are pending full-text review.</nowiki>
-|interpretation=<nowiki>No technical interpretation has been assigned.</nowiki>
+For very small payloads, PRP setup and PCIe data movement dominate useful transfer work.
-|reusable_lesson=<nowiki>Pending full-text review.</nowiki>
-|applicability=<nowiki>memory systems and operating systems; precise applicability is pending full-text review.</nowiki>
+Verification: official DOI metadata and author publication page/abstract; confidence=high.</nowiki>
-|confidence=<nowiki>high</nowiki>
+|observation=<nowiki>workloads=sub-page NVMe payload microbenchmarks on Linux/OpenSSD; baselines=PRP and prior small-I/O method; metrics=PCIe traffic and latency; results=up to 98% less traffic; 40% and 39% lower latency</nowiki>
+|interpretation=<nowiki>Command transport can double as a data path when payloads are smaller than conventional DMA setup costs.</nowiki>
+|reusable_lesson=<nowiki>Inline tiny values into control messages when protocol headroom and ordering permit.</nowiki>
+|applicability=<nowiki>NVMe workloads dominated by small reads/writes on modifiable controllers.
+Limits: Prototype uses OpenSSD and targets sub-page payloads; benefit does not generalize to large transfers.</nowiki>
+|confidence=<nowiki>medium</nowiki>
 |evidence=<nowiki>ByteExpress: A High-Performance and Traffic-Efficient Inline Transfer of Small Payloads over NVMe. HotStorage 2025.</nowiki>
 |record_origin=<nowiki>lab</nowiki>
@@ 15번째 줄: / 21번째 줄: @@
 |review_state=<nowiki>Draft</nowiki>
 |created_at=<nowiki>2026-07-16T14:58:56.771380Z</nowiki>
-|updated_at=<nowiki>2026-07-16T14:59:04.408468Z</nowiki>
+|updated_at=<nowiki>2026-07-18T14:58:21.143910Z</nowiki>
 }}
@@ 26번째 줄: / 32번째 줄: @@
 |added_by=<nowiki>S3ResearchAgent</nowiki>
 |added_at=<nowiki>2026-07-16T14:59:04.408468Z</nowiki>
+}}
+{{Lesson evidence
+|id=<nowiki>verified-content-v1-0115</nowiki>
+|citation=<nowiki>Junhyeok Park et al., "ByteExpress: Inline Small-Payload Transfer for NVMe", HotStorage 2025.</nowiki>
+|url=<nowiki>https://doi.org/10.1145/3736548.3737837</nowiki>
+|kind=<nowiki>paper</nowiki>
+|note=<nowiki>Verification: official DOI metadata and author publication page/abstract; confidence=high.
+Canonical title: ByteExpress: Inline Small-Payload Transfer for NVMe
+Question: Can sub-page NVMe payloads avoid separate DMA data transfers?
+Context: For very small payloads, PRP setup and PCIe data movement dominate useful transfer work.
+Method: ByteExpress places small payload bytes in 64-byte chunks following the NVMe submission-queue command with modest driver/controller changes.
+Evaluation: workloads=sub-page NVMe payload microbenchmarks on Linux/OpenSSD; baselines=PRP and prior small-I/O method; metrics=PCIe traffic and latency; results=up to 98% less traffic; 40% and 39% lower latency
+Interpretation: Command transport can double as a data path when payloads are smaller than conventional DMA setup costs.
+Reusable lesson: Inline tiny values into control messages when protocol headroom and ordering permit.
+Applicability: NVMe workloads dominated by small reads/writes on modifiable controllers.
+Limits: Prototype uses OpenSSD and targets sub-page payloads; benefit does not generalize to large transfers.</nowiki>
+|added_by=<nowiki>S3ResearchAgent</nowiki>
+|added_at=<nowiki>2026-07-16T18:44:31.574353Z</nowiki>
+}}
+{{Lesson evidence
+|id=<nowiki>canonical-paper-v2-a7c01a49</nowiki>
+|citation=<nowiki>Junhyeok Park et al., "ByteExpress: A High-Performance and Traffic-Efficient Inline Transfer of Small Payloads over NVMe", HotStorage 2025.</nowiki>
+|url=<nowiki>https://doi.org/10.1145/3736548.3737837</nowiki>
+|kind=<nowiki>paper</nowiki>
+|verification_basis=<nowiki>official_abstract</nowiki>
+|note=<nowiki>Verification: official DOI metadata and author publication page/abstract; confidence=medium.
+Canonical title: ByteExpress: A High-Performance and Traffic-Efficient Inline Transfer of Small Payloads over NVMe
+Question: Can sub-page NVMe payloads avoid separate DMA data transfers?
+Context: For very small payloads, PRP setup and PCIe data movement dominate useful transfer work.
+Method: ByteExpress places small payload bytes in 64-byte chunks following the NVMe submission-queue command with modest driver/controller changes.
+Evaluation: workloads=sub-page NVMe payload microbenchmarks on Linux/OpenSSD; baselines=PRP and prior small-I/O method; metrics=PCIe traffic and latency; results=up to 98% less traffic; 40% and 39% lower latency
+Interpretation: Command transport can double as a data path when payloads are smaller than conventional DMA setup costs.
+Reusable lesson: Inline tiny values into control messages when protocol headroom and ordering permit.
+Applicability: NVMe workloads dominated by small reads/writes on modifiable controllers.
+Limits: Prototype uses OpenSSD and targets sub-page payloads; benefit does not generalize to large transfers.</nowiki>
+|added_by=<nowiki>S3ResearchAgent</nowiki>
+|added_at=<nowiki>2026-07-18T05:13:51.779149Z</nowiki>
+}}
+{{Lesson evidence verification
+|id=<nowiki>verify_57574e60c1ac63039638</nowiki>
+|evidence_id=<nowiki>ev_686c9b51ad8f434d</nowiki>
+|evidence_digest=<nowiki>012e2eac713df4c37e6a3810c3eb4d967c5f9e31ba2cd8c500665ad7fec92504</nowiki>
+|verification_basis=<nowiki>metadata_only</nowiki>
+|source_identity=<nowiki>R2-RESTIC:7f893ca5afd2cfb6fe320e9b61063ccc70e75a7a96589420038c8cf338b273be; archive-manifest-sha256=e28171fb69e141ce306d92dfe4b10e6cdc6e81d4fa910c30a846204dbcf8edf8; sha256=487c7ad325b2def210ccc911954681c7c864870247d6c7488f1061f256729363</nowiki>
+|source_sha256=<nowiki>487c7ad325b2def210ccc911954681c7c864870247d6c7488f1061f256729363</nowiki>
+|source_locator=<nowiki>보존 파일 objects/sha256/48/487c7ad325b2def210ccc911954681c7c864870247d6c7488f1061f256729363</nowiki>
+|coverage=<nowiki>보존 객체는 cookie/landing page이므로 서지 위치만 확인했고 본문 주장을 검증하지 못함.</nowiki>
+|outcome=<nowiki>inconclusive</nowiki>
+|claim_fields=<nowiki>context</nowiki>
+|verified_by=<nowiki>S3ResearchAgent</nowiki>
+|verified_at=<nowiki>2026-07-18T14:58:20.716830Z</nowiki>
+}}
+{{Lesson evidence verification
+|id=<nowiki>verify_cd53e771a9c9105dae11</nowiki>
+|evidence_id=<nowiki>verified-content-v1-0115</nowiki>
+|evidence_digest=<nowiki>81e1be00a0e8033bd2e514a783ae5e056c0ed309c3e2a94c20e35da6972b89a5</nowiki>
+|verification_basis=<nowiki>metadata_only</nowiki>
+|source_identity=<nowiki>R2-RESTIC:7f893ca5afd2cfb6fe320e9b61063ccc70e75a7a96589420038c8cf338b273be; archive-manifest-sha256=e28171fb69e141ce306d92dfe4b10e6cdc6e81d4fa910c30a846204dbcf8edf8; sha256=a71b309299481eade86b099e9831dc07c19f85a095f939e2b06f75f38fc270ea</nowiki>
+|source_sha256=<nowiki>a71b309299481eade86b099e9831dc07c19f85a095f939e2b06f75f38fc270ea</nowiki>
+|source_locator=<nowiki>보존 파일 objects/sha256/a7/a71b309299481eade86b099e9831dc07c19f85a095f939e2b06f75f38fc270ea</nowiki>
+|coverage=<nowiki>보존 객체는 cookie/landing page이므로 서지 위치만 확인했고 본문 주장을 검증하지 못함.</nowiki>
+|outcome=<nowiki>inconclusive</nowiki>
+|claim_fields=<nowiki>context</nowiki>
+|verified_by=<nowiki>S3ResearchAgent</nowiki>
+|verified_at=<nowiki>2026-07-18T14:58:20.979258Z</nowiki>
+}}
+{{Lesson evidence verification
+|id=<nowiki>verify_e580b8c07c1188c71965</nowiki>
+|evidence_id=<nowiki>canonical-paper-v2-a7c01a49</nowiki>
+|evidence_digest=<nowiki>53326b876ea1cfb2c9eb0395d6965df48307024ecb35479012754f03a47bcb7f</nowiki>
+|verification_basis=<nowiki>metadata_only</nowiki>
+|source_identity=<nowiki>R2-RESTIC:7f893ca5afd2cfb6fe320e9b61063ccc70e75a7a96589420038c8cf338b273be; archive-manifest-sha256=e28171fb69e141ce306d92dfe4b10e6cdc6e81d4fa910c30a846204dbcf8edf8; sha256=a71b309299481eade86b099e9831dc07c19f85a095f939e2b06f75f38fc270ea</nowiki>
+|source_sha256=<nowiki>a71b309299481eade86b099e9831dc07c19f85a095f939e2b06f75f38fc270ea</nowiki>
+|source_locator=<nowiki>보존 파일 objects/sha256/a7/a71b309299481eade86b099e9831dc07c19f85a095f939e2b06f75f38fc270ea</nowiki>
+|coverage=<nowiki>보존 객체는 cookie/landing page이므로 서지 위치만 확인했고 본문 주장을 검증하지 못함.</nowiki>
+|outcome=<nowiki>inconclusive</nowiki>
+|claim_fields=<nowiki>context</nowiki>
+|verified_by=<nowiki>S3ResearchAgent</nowiki>
+|verified_at=<nowiki>2026-07-18T14:58:21.143910Z</nowiki>
 }}