盤點分佈式軟總線數據傳輸技術中的黑科技｜HDC2021技術分論壇

{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}},{"type":"color","attrs":{"color":"#40A9FF","name":"blue"}}],"text":"作者：houweibo，軟總線首席技術專家；lidonghua，軟總線技術專家","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"隨着萬物互聯時代的到來，特別是大量媒體資源的湧入和使用，用戶對傳輸的要求不斷提高，怎樣的傳輸技術才能滿足未來的用戶需求呢？ ","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在萬物互聯、多設備協同的新場景下，業界通用的OSI模型在滿足用戶對傳輸的高要求方面顯得力有未逮。尋求一種極致的傳輸技術勢在必行，這也成爲分佈式軟總線的一項重要任務。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"下面讓我們先從“超級終端”新場景視角分析OSI模型，然後一一介紹分佈式軟總線的數據傳輸技術中的黑科技。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"一、“超級終端”新場景視角看OSI模型","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"開放系統互聯通信參考模型（Open System Interconnection Reference Model，簡稱OSI模型）是一個試圖使世界範圍各種計算機互聯爲網絡的標準框架。OSI模型並沒有提供一個可以實現的方法，而是描述了一些概念，用來協調進程間通信標準的制定。這一模型定義了網絡互連的七層框架（物理層、數據鏈路層、網絡層、傳輸層、會話層、表示層和應用層）。通常也將會話層、表示層和應用層合併爲一層，統稱爲應用層。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"OSI模型適用於將不同型號的計算機互連成一個單一的網絡，它極大地推動了網絡標準化的進程。然而隨着鴻蒙生態的急速拓展，越來越多不同類型設備連接在一起形成“超級終端”， OSI模型已無法支持這樣新場景下的數據傳輸要求。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"比如生活中，我們經常看到Wi-Fi版本標稱最大可達幾百至上千兆帶寬，這麼高的帶寬真的是用戶實際使用帶寬嗎？事實上，Wi-Fi的最大帶寬和用戶使用帶寬是兩回事，這其中也隱含着以OSI爲原型的網絡模型的些許不足之處。","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/f8/f81de4222da83c55bc146de44bd2e335.webp","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"圖1 超級終端新場景下OSI模型","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在應用層，設備接入數量、接入信道以及信號強度和干擾等均會影響傳輸效率。在應用層以下，各層協議數據存在大量數據開銷，同時MAC報文傳輸過程耗時也較大。在物理層，器件規格和協議規格也直接影響着帶寬的理論上限。這些問題從本質上是沒有對時間和空間資源進行高效利用，沒有發揮資源的最大化使用價值。如何解決現有模型的不足之處，實現時間和空間等資源的最大化利用，成爲新一代數據傳輸技術的重要突破點。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"二、分佈式軟總線數據傳輸技術中的黑科技","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"與傳統P2P（point-to-point，點對點）傳輸有着較大不同，多設備場景下資源可調度、可利用空間更加廣闊。朝着“超級終端”多設備資源最大化利用的目標，分佈式軟總線數據傳輸技術推出了三項黑科技，鑄造極致的傳輸能力和體驗。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"1. 多徑傳輸技術（Multi-Path Transport）","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"近年來，高清音視頻、動畫等多媒體資源日益增長，實現通用高效率文件傳輸成爲一項極其重要的需求。然而傳統的文件傳輸，往往依賴一條鏈路直接進行傳輸。文件的傳輸效率完全依賴鏈路本身的傳輸能力，魯棒性和傳輸效率均不足。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"爲了適應新互聯時代的需求，實現高效率文件傳輸，分佈式軟總線多徑傳輸技術應運而生。多徑傳輸技術基本思路是最大化利用不同路徑的傳輸能力，原本一條鏈路上的文件傳輸，通過文件分割，將分割後的文件不同部分從多條路徑進行傳輸，最終在目標側聚合成原有文件。通過智能均衡不同鏈路的傳輸負載，實現了對文件的分割和傳輸，能力強的鏈路多分些文件佔比，能力弱的少分些文件佔比，甚至可以不分。這樣就實現了文件的高效快速傳輸，給用戶帶來不一樣的快速傳輸體驗。","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/0b/0b3c7fce89089cf6a74a7e37e292cd07.webp","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"圖2 多路徑傳輸文件","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"2. 智能時延控制技術（Intelligent Latency Control）","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在網絡傳輸過程中，很大一塊開銷是序列化和協議編碼佔據的。進一步優化這部分開銷，對傳輸時延和空間資源有着重要意義。智能時延控制技術正是通過智能序列化技術和極簡傳輸協議兩項重要子技術，實現了序列化和協議編碼的極大壓縮。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"（1）智能序列化技術（Intelligent Serialization）","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在RPC（remote procedure call，遠程過程調用）通信流程中，百KB級別的RPC消息序列化需要計算側耗時4~5ms，網絡側耗時約10~15ms，系統加速比低，嚴重影響了系統的運行性能。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"智能序列化技術通過對計算側、網絡側和能耗側進行綜合建模，通過最優化方法和智能決策策略實現了智能的序列化方案，使計算側和網絡側總時延降至8ms，縮減60%+。 ","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/89/899c7fffe2f947d21bb4d100751cd549.webp","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"圖3 智能序列化技術","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"（2）極簡傳輸協議（Minimalist Transport Protocol）","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"隨着文件及流媒體文件對傳輸效率和速度提出更高的要求，用戶對全場景傳輸效率的需求進一步提升，正是爲了適應這樣的新要求，極簡傳輸協議應運而生。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"極簡傳輸協議，將OSI模型中表示層、會話層、傳輸層和網絡層協議棧精簡爲一層，通過報文簡化、包頭簡化、交互簡化，場景感知的緩衝機制等方式，提升有效的傳輸荷載、突破 TCP/IP 協議的傳輸限制，帶來了傳輸效率與速度的跨越式提升，是分佈式軟總線的核心技術之一。相較傳統協議，時延縮減70%+。","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/bd/bd89549128bcbf2395d70b95252f693a.webp","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"圖4 極簡傳輸協議","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"3. 高效動態資源調度技術（Dynamic Resource Scheduling）","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"多設備協同場景下，設備間競爭、干擾和衝突避讓等機制嚴重限制了空口的原有能力，帶寬急劇降低，同時時延不可預期。就這個問題，軟總線團隊在多設備場景上長期深耕，推出了高效動態資源調度技術。","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/e2/e20374101590bb26de692c4709ef50b1.webp","alt":null,"title":"","style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"size","attrs":{"size":10}}],"text":"圖5 多設備協同場景","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"高效動態資源調度技術包括以下三項子技術： ","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"（1）秩序化組網技術","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"秩序化組網，顧名思義，這是一種確定性網絡技術。秩序化組網技術通過秩序化自組織網絡的秩序化特性，將原本“無序”的網絡變得“有序”，通過有序可控的調度實現空口資源最優化利用，從而帶來空口利用率的極大提升，實現了空口資源利用的最優調度策略。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"（2）“超級終端”拓撲數傳技術","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"用戶看到某個設備便可以操控和使用它（“所見即所用”）一直是用戶的深層渴望。受傳統的P2P角色限制，物聯網設備間並不能實現真正意義的“所見即所用”，這極大限制了資源的利用，嚴重影響了互聯背景下的用戶體驗。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"“超級終端”拓撲數傳技術，通過建立HML私有協議，使超級終端內部設備都成爲可支配的計算和傳輸資源，實現了“所見即所用”模式。設備之間的隔閡漸趨消失，滿足了用戶對設備使用體驗的深層渴望。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"（3）“超級終端”智能調度技術","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"超級終端內不同設備的通信和計算能力不同，如何合理配置這些資源，實現“能者多勞”和發揮“一技之長”。讓通信能力強的資源用在通信上，計算能力強或專業計算的資源幫助計算劣勢的資源完成複雜高負荷的計算，成爲調度技術的主要關注問題。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在秩序化組網的有序性基礎上，“超級終端”智能調度技術，通過軟件SDK和CPU的智能調度策略和算法，實現了傳輸和計算資源的合理分配，最大發揮“能力強”和“專業硬”的設備通信和計算能力。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"三、總結與展望","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在萬物互聯時代，如何極致化利用資源，最大化發揮資源能力和價值，一直是分佈式軟總線數據傳輸技術的探索思路和目標。未來數據傳輸技術還會持續挖掘新場景下更加高效快速、更加可靠的傳輸能力，持續爲超級終端、爲物聯網世界注入更強大的傳輸技術和能力。","attrs":{}}]}]}