探訪華爲雲全球最大雲數據中心，背後藏着這些“黑科技”

{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"新基建背景下，數據中心作爲支撐新基建發展的重要IT基礎設施，愈發受到重視。除三大運營商外，BAT等互聯網巨頭近年也開始大力投入數據中心的建設和佈局。近日，InfoQ記者來到華爲雲貴安數據中心實地探訪，進一步瞭解AI和大數據等技術在支撐超大型數據中心日常運轉所發揮的作用。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.geekbang.org\/infoq\/97\/979d8b51636fbebf7a9828c4e6f6ee7a.jpeg","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":"center","origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#a5a5a5","name":"user"}}],"text":"坐落在貴安新區數谷大道1號的“童話小鎮”，也是華爲雲全球最大雲數據中心"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"2016年，華爲與貴州省政府簽署戰略合作協議，數據中心正式投建。目前華爲雲貴安數據中心一期已經投入使用，共建設有9棟機房，預計未來三到五年還會有更多機房建成。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"據華爲雲營銷部長董理斌介紹，貴安數據中心規劃爲華爲全球最大的雲數據中心，全部建成後可容納100萬臺服務器。它也是華爲雲業務的重要承載節點，主要承載華爲雲、消費者雲和華爲內部流程IT等業務。“如果以貴州爲中心，用一千公里畫一個半徑，貴安華爲雲數據中心的服務範圍能夠輻射到重慶、廣西、廣東、雲南、四川等周邊省份和地區。”"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"除了建設數據中心以外，華爲雲貴安數據中心還將承擔華爲全球IT維護工程師基地、員工培訓實習基地的職能。預計將有約600-800位IT維護工程師對數據中心提供支持與服務，每年還將有大量人員到園區進行全景化實戰培訓、實習等。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"當前，華爲雲在中國佈局了五大數據中心，除了貴安和烏蘭察布外，還有京津冀、長三角、粵港澳片區三大核心數據中心。在國內數據中心規劃中，華爲雲主要基於時延來進行數據中心的冷、溫、熱佈局，其中冷服務主要建在低成本地方，溫服務貼近沿海的低成本地方，熱服務則佈局在貼近客戶需求的地方。在海外，華爲也在歐洲、中東、非洲、亞太、拉美等區域建立了本地數據中心。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"在董理斌看來，當前數據中心及相關聯產業目前仍處在飛速發展階段。僅在貴州，華爲雲就已爲超過800家貴州企業數字化轉型提供服務，全省62家省直部門1438個數據資源都已上雲。而據中國信息通信研究院數據，截至2020年底，我國在用數據中心機架總規模超過400萬架，近5年年均增速超過30%。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"但數據中心產業快速發展的同時，也帶來了能耗大幅增長的問題。據《中國數據中心能耗現狀白皮書》，早在2015年，全國大數據中心的耗電量已達1000億kWh，相當於三峽電站全年的發電量；2018年這個數值迅速爬升至1609億kWh，超過上海全年的社會用電量。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"能耗問題如何破解？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"今年7月14日，工業和信息化部印發《新型數據中心發展三年行動計劃（2021-2023年）》（以下簡稱“行動計劃”）明確指出：到2021年底，全國數據中心平均利用率力爭提升到55%以上，總算力超過120EFLOPS，新建大型及以上數據中心PUE降低到1.35以下；到2023年底，全國數據中心機架規模年均增速保持在20%左右，平均利用率力爭提升到60%以上，新建大型及以上數據中心PUE降低到1.3 以下，嚴寒和寒冷地區力爭降低到1.25以下。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"PUE，即Power Usage Effectiveness，這是一項用於評價數據中心能源效率的指標，它的值取自數據中心消耗的所有能源（總能耗）與IT設備消耗的能源的比值。 其中數據中心總能耗包括IT設備能耗和製冷、配電等系統的能耗，因此PUE值通常大於1，PUE值越接近1表明非IT設備耗能越少，即能效水平越好。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"華爲雲貴安數據中心的能效比PUE做到了1.12，處於業界領先水平。董理斌在採訪中向InfoQ表示，爲了把PUE做到儘可能低，華爲雲採取了多重手段。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"首先，數據中心選址貴安主要也是基於氣候的考慮，貴安年均溫度約15℃，空氣質量良好，沒有化學、顆粒物污染，使採用直通風自然冷卻成爲可能。爲了充分利用自然冷源，放置服務器的機房建築相比其他普通建築也做了一些特殊設計，包括更高的層高和用百葉窗替代普通窗戶，而百葉窗背後其實就是一個個的機櫃。冷風經大樓百葉窗送進機房，熱風經熱通道從樓頂排出。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.geekbang.org\/infoq\/7c\/7c33638c8e18bd8339727674a4ca6ae8.jpeg","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" 其次，數據中心設施區和辦公區域採用中溫冷凍水系統，散發的熱量一部分通過瀑布和湖面實現自然冷卻，另一個部分通過熱回收技術用於游泳池和辦公區冬季制熱，實現餘熱回收，減少熱損耗。此外還引入了服務器液冷技術，能夠把熱量直接從芯片上帶走，減少製冷無關損耗，提升設備密度降低能效比。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"AI技術也在其中發揮了重要作用，基於AI技術能夠實現隨業務功率的變化實時調整製冷功率、削峯平谷，使各服務器負荷均衡，進一步提升資源使用效率比。同時在供電環節用功率半導體替換銅器件，結合智能算法，又能進一步降低供電損耗。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"華爲雲稱，在滿負荷運行的情況下，預計每年可節省電力10.1億度，減少碳排放81萬噸，相當於年植樹3567萬棵。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"智能運維：人均運維設備數量提升10倍"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"在數據中心生命週期中，數據中心運維管理是數據中心生命週期中最後一個、也是歷時最長的一個階段。如何用盡可能少的運維人員高效完成數據中心的運維工作，同樣是一個非常大的挑戰。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"隨着數據中心規模不斷擴大，當故障發生時，想要在幾十萬甚至上百萬臺服務器中，通過人工來確定故障具體發生在哪個機房、哪個機櫃、哪臺機器，並找到出問題的具體原因，已經非常困難。爲了解決這個問題，華爲雲結合智能傳感和IoT等技術構建了數字孿生平臺，可以實時顯示數據中心環境和設備的狀態，把供電和製冷全鏈路狀態以及各模塊每臺服務器工作狀態、工作溫度全部可視化。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/ef\/e0\/efeb6cce500e094cda6c65482bde08e0.jpg","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":"color","attrs":{"color":"#494949","name":"user"}}],"text":"運行中的河圖機器人"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.infoq.cn\/resource\/image\/f7\/03\/f7b6879e8662842e2d39849f13cd6303.jpg","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":"color","attrs":{"color":"#494949","name":"user"}}],"text":"機房環境信息實時展示"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"結合華爲河圖虛擬增強現實技術，運維工程師可以非常方便地實時看到機櫃溫度、溼度、通風等機房環境信息，人均運維設備數量提升了10倍，故障定界定位時間也從30分鐘大幅降低到1分鐘。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"據瞭解，河圖是華爲在2019年8月推出的底層技術平臺，定位數據基礎設施，其核心能力包括全場景空間計算能力、AR步行導航、場景編輯、渲染等核心技術。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"此外，藉助大數據分析和AI技術，還可以提前分析可能存在的隱患，對基礎設施故障進行預測併發布智能風險提醒，從而把運維工作從“被動”變爲“主動”，防範於未然。"}]}]}