[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"article-data-link-layer-osi-layer-2-en-zh":3,"article-related-data-link-layer-osi-layer-2-en-zh":31,"series-research-1eed455b-155a-4dce-9aaf-7f8328e9614f":76},{"id":4,"slug":5,"title":6,"content":7,"summary":8,"source":9,"source_url":10,"author":11,"image_url":12,"cover_image":12,"category":13,"language":14,"translated_content":11,"related_article_id":15,"keywords":16,"key_takeaways":23,"views":27,"created_at":28,"published_at":29,"topic_cluster_id":30},"1eed455b-155a-4dce-9aaf-7f8328e9614f","data-link-layer-osi-layer-2-en-zh","OSI 第2層：資料鏈結層速懂","\u003Cp>兩台設備接在同一個網段，資料還是不能隨便送。誰先講、封包怎麼包成 frame、出錯後怎麼處理，這些都屬於 OSI 第 2 層。\u003C\u002Fp>\u003Cp data-speakable=\"summary\">資料鏈結層負責同一網段內的 frame 傳輸、MAC 位址與媒體存取。\u003C\u002Fp>\u003Cp>這次重點不是新理論，而是把 \u003Ca href=\"\u002Ftag\u002Flayer-2\">layer 2\u003C\u002Fa> 的位置、分工和常見實作講清楚。對做網路、寫系統或在現場排障的\u003Ca href=\"\u002Fnews\u002Fmistral-leanstral-proof-engineering-open-model-zh\">工程\u003C\u002Fa>師來說，這一層常常就是問題開始冒頭的地方。\u003C\u002Fp>\u003Ch2>發生了什麼\u003C\u002Fh2>\u003Cp>資料鏈結層位在實體層和網路層之間，處理的是節點到節點的本地傳輸。它不負責跨網段路由，而是把資料整理成 frame，讓同一條鏈路或同一個 LAN 上的裝置能互相溝通。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1783348374154-2340.png\" alt=\"OSI 第2層：資料鏈結層速懂\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>在 OSI \u003Ca href=\"\u002Fnews\u002Fosi-model-explained-networking-layers-zh\">模型\u003C\u002Fa>裡，layer 2 常再拆成兩個子層：LLC 與 MAC。LLC 負責多工上層協定，也可能提供 flow control 或 acknowledgment；MAC 則決定誰能使用媒體，以及 frame 何時送出、怎麼同步、怎麼定址。\u003C\u002Fp>\u003Cul>\u003Cli>Layer 2 傳的是 frame，不是 packet。\u003C\u002Fli>\u003Cli>Frame header 會帶 source 和 destination MAC address。\u003C\u002Fli>\u003Cli>很多 LAN 用的是平面位址，和可路由的網路層位址不同。\u003C\u002Fli>\u003Cli>Ethernet 常見 CSMA\u002FCD，Wi‑Fi 常見 CSMA\u002FCA。\u003C\u002Fli>\u003Cli>有些協定會加上錯誤偵測、重傳或 flow control。\u003C\u002Fli>\u003C\u002Ful>\u003Cp>不同標準對 layer 2 的切法也不完全一樣。IEEE 802 系列多半把 LLC 和 MAC 分開寫，像 HDLC 這類協定則可能把兩者合在一起；家用電力線網路的 G.hn 還會再加一層 convergence sublayer，處理不同上層流量的銜接。\u003C\u002Fp>\u003Cp>如果把它放回日常網路來看，layer 2 做的其實是本地交通管理。它先確認資料能不能在這條鏈路上安全送出，再把後續的路由工作交給第 3 層。\u003C\u002Fp>\u003Ch2>為什麼重要\u003C\u002Fh2>\u003Cp>對開發者來說，layer 2 會直接影響裝置能不能在局部網路裡正常說話。交換器轉發、VLAN、MAC filtering、碰撞處理和 frame 邊界，這些都不是應用程式層能補救的事。\u003C\u002Fp>\u003Cp>這也解釋了為什麼有些故障只發生在本地，有些則會一路擴散。Ethernet 在有線短距離環境裡通常不需要在 layer 2 做太多重傳，但無線或某些調變式連線，常得靠 link-layer recovery 來撐住錯誤率。\u003C\u002Fp>\u003Cp>實務上，這會影響延遲、吞吐量，還有你應該把錯誤處理寫在哪一層。當封包看起來像「送不出去」，先查的往往不是 \u003Ca href=\"\u002Ftag\u002Fapi\">API\u003C\u002Fa>，而是交換器設定、MAC 位址表、媒體存取規則，或是實體連線本身。\u003C\u002Fp>\u003Cp>TCP\u002FIP 模型把這些工作收進 link layer，但 OSI 的說法仍然好用，因為它能快速把問題定位到正確區段。對除錯來說，這種分層不是學術名詞，而是縮短排查路徑的方法。\u003C\u002Fp>\u003Cp>換句話說，layer 2 是網路的本地規則書。資料還沒離開這個區段之前，很多成敗就已經決定了。\u003C\u002Fp>\u003Cp>如果你在看 packet loss、Wi‑Fi 不穩，或交換器上的異常流量，先問的通常不是「應用壞了沒」，而是「第 2 層到底有沒有正常轉起來」。\u003C\u002Fp>\u003Ch2>補充背景\u003C\u002Fh2>\u003Cp>OSI 模型把網路問題切成七層，是為了讓不同團隊能用同一套語言溝通。這套語言在現場排障時特別有用，因為它能把「裝置看得到、資料送不動」和「跨網段不通」分開看。\u003C\u002Fp>\n\u003Cfigure class=\"my-6\">\u003Cimg src=\"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1783348383611-x1h9.png\" alt=\"OSI 第2層：資料鏈結層速懂\" class=\"rounded-xl w-full\" loading=\"lazy\" \u002F>\u003C\u002Ffigure>\n\u003Cp>和第 3 層相比，layer 2 的範圍更小，但細節更密。第 3 層看的是可路由位址與跨網段傳遞，第 2 層看的是同一段鏈路內的接入、定址與資料封裝，兩者常常一起出問題，但責任邊界不同。\u003C\u002Fp>\u003Cp>這也是為什麼交換器、網卡驅動、無線接入點和 VLAN 設定，會在很多事故裡成為第一批檢查項目。只要本地 frame 不穩，後面的 DNS、API、\u003Ca href=\"\u002Fnews\u002Fbuild-production-vector-db-rag-pipeline-zh\">資料庫\u003C\u002Fa>連線都可能看起來像壞掉，其實只是底層傳輸失常。\u003C\u002Fp>\u003Cp>對網路工程師而言，理解 layer 2 不是背名詞，而是知道哪個層級能解哪種錯。對應用開發者來說，它則是提醒：有些延遲和丟包，真的不是程式碼能單獨解決的。\u003C\u002Fp>\u003Cp>真正的問題通常很樸素：資料有沒有在同一條鏈路上被正確接收。這就是第 2 層的工作，也是它最常被忽略、卻最容易出事的地方。\u003C\u002Fp>","資料鏈結層負責同一網段內的 frame 傳輸、MAC 位址與媒體存取，常見於 Ethernet、Wi‑Fi 與交換器除錯。","en.wikipedia.org","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FData_link_layer",null,"https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1783348374154-2340.png","research","zh","7fdb9c93-55a6-4f1a-968a-6e683b200191",[17,18,19,20,21,22],"OSI","data link layer","layer 2","MAC address","Ethernet","Wi-Fi",[24,25,26],"Layer 2 負責同一網段內的 frame 傳輸，不做跨網段路由。","LLC 與 MAC 是理解資料鏈結層分工的兩個核心子層。","除錯時，交換器、VLAN、MAC 表與媒體存取規則常是第一批檢查點。",0,"2026-07-06T14:32:28.437901+00:00","2026-07-06T14:32:28.423+00:00","772f04d1-a22a-4aba-9926-8fb34d565cbe",{"tags":32,"relatedLang":35,"relatedPosts":39},[33],{"name":19,"slug":34},"layer-2",{"id":15,"slug":36,"title":37,"language":38},"data-link-layer-osi-layer-2-en","Data link layer: OSI layer 2 explained","en",[40,46,52,58,64,70],{"id":41,"slug":42,"title":43,"cover_image":44,"image_url":44,"created_at":45,"category":13},"ccf82d31-981a-47a5-b5e2-970ee982b11e","camvla-calibration-free-view-robust-vla-zh","CamVLA 讓機器人不怕換鏡頭","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1783404180796-s1pg.png","2026-07-07T06:02:31.413058+00:00",{"id":47,"slug":48,"title":49,"cover_image":50,"image_url":50,"created_at":51,"category":13},"f64c8577-f362-4ef5-b1ac-78859c83ed26","osi-model-explained-networking-layers-zh","OSI 模型到現在還好用","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1783346588623-i7pm.png","2026-07-06T14:02:38.037796+00:00",{"id":53,"slug":54,"title":55,"cover_image":56,"image_url":56,"created_at":57,"category":13},"404bac33-b9b4-41bb-bb9a-1d98a63aa536","evaluation-protocols-fine-tuned-llms-2026-zh","2026 微調 LLM 評估流程","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1783101776530-b8eu.png","2026-07-03T18:02:24.572198+00:00",{"id":59,"slug":60,"title":61,"cover_image":62,"image_url":62,"created_at":63,"category":13},"8f3122c8-9eb1-4aa6-b780-3b62003b3418","deepspec-data-regeneration-pipeline-qwen3-eagle3-zh","DeepSpec 應被視為資料重生管線，而不是訓練技巧","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1783080165006-321z.png","2026-07-03T12:02:18.375863+00:00",{"id":65,"slug":66,"title":67,"cover_image":68,"image_url":68,"created_at":69,"category":13},"6cfddc0d-ce6e-4a14-baf7-3531bf32bc5d","program-as-weights-fuzzy-functions-zh","PAW把提示詞編成可重用工具","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1783062178440-pnt0.png","2026-07-03T07:02:32.5878+00:00",{"id":71,"slug":72,"title":73,"cover_image":74,"image_url":74,"created_at":75,"category":13},"5bd0dc27-5a7f-4563-8086-acccc98eb2fc","lacuna-llm-unlearning-localization-testbed-zh","LACUNA：檢驗 LLM 真的有沒有忘記","https:\u002F\u002Fxxdpdyhzhpamafnrdkyq.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Fcovers\u002Finline-1783060373883-d92j.png","2026-07-03T06:32:31.28626+00:00",[77,82,87,92,97,102,107,112,117,122],{"id":78,"slug":79,"title":80,"created_at":81},"f18dbadb-8c59-4723-84a4-6ad22746c77a","deepmind-bets-on-continuous-learning-ai-2026-zh","DeepMind 押注 2026 連續學習 AI","2026-03-26T08:16:02.367355+00:00",{"id":83,"slug":84,"title":85,"created_at":86},"f4a106cb-02a6-4508-8f39-9720a0a93cee","ml-papers-of-the-week-github-research-desk-zh","每週 ML 論文清單，為何紅到 GitHub","2026-03-27T01:11:39.284175+00:00",{"id":88,"slug":89,"title":90,"created_at":91},"c4f807ca-4e5f-47f1-a48c-961cf3fc44dc","ai-ml-conferences-to-watch-in-2026-zh","2026 AI 研討會投稿時程整理","2026-03-27T01:51:53.874432+00:00",{"id":93,"slug":94,"title":95,"created_at":96},"cf046742-efb2-4753-aef9-caed5da5e32e","adaptive-block-scaled-data-types-zh","IF4：神經網路量化的聰明選擇","2026-03-31T06:00:36.990273+00:00",{"id":98,"slug":99,"title":100,"created_at":101},"53a0dc54-0371-4e40-8d5e-74e94a73840c","geometry-aware-similarity-metrics-for-neural-representations-zh","超越距離測量：用微分幾何重新理解神經網路","2026-03-31T06:01:01.241968+00:00",{"id":103,"slug":104,"title":105,"created_at":106},"fee7d472-a775-4b1d-bbc2-1e8bca1bbf8b","on-the-fly-repulsion-in-the-contextual-space-for-rich-divers-zh","讓AI繪圖更有創意：用排斥力提升生成多樣性","2026-03-31T06:01:25.439673+00:00",{"id":108,"slug":109,"title":110,"created_at":111},"a9901203-d69b-447b-8854-15d14eab32b4","vision-aided-beam-prediction-cnn-eca-zh","影像輔助波束預測升級 CNN","2026-04-01T10:00:25.8073+00:00",{"id":113,"slug":114,"title":115,"created_at":116},"b55e7dd4-0a24-4b3d-804d-b0309a03f498","triple-band-fss-mimo-antenna-sub-6-ghz-zh","三頻 FSS MIMO 天線瞄準 sub-6 GHz","2026-04-01T13:18:36.857305+00:00",{"id":118,"slug":119,"title":120,"created_at":121},"f68290bd-e7f3-4b30-ba22-dcd4e0130a66","openclaw-1299-repos-eight-weeks-analysis-zh","OpenClaw 1299 個 Repo 的資料解讀","2026-04-02T05:03:45.208411+00:00",{"id":123,"slug":124,"title":125,"created_at":126},"ed9f80eb-eb02-4d35-8ad4-0ddf428751dd","beam-coherence-aware-combining-mmwave-mimo-zh","毫米波 MIMO 的雙階合併法","2026-04-02T05:27:26.897188+00:00"]