Databricks與Snowflake創始人開撕：“未來十年數據倉庫要麼不存在要麼大變樣”

{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Databricks 與 Snowflake 之間的激烈競爭再上新臺階，甚至有可能給整個數據倉庫領域帶來更加深遠的影響。"}]}]},{"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","text":"短短半個月，大數據領域新一代領軍企業 Databricks 和 Snowflake 就互撕了幾回。"}]},{"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","text":"11 月 2 日，Databricks 在其官方博客發佈聲明，表示其數據湖倉（lake house）技術創下 TPC-DS 基準測試新記錄，並強調第三方研究表明實際性能可達 Snowflake 的 2.5 倍。"}]},{"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","text":"在博客中，Databricks 聲稱這是一件大事，有助於證明數據倉庫在未來十年要麼不復存在，要麼會大變樣，“從長遠來看，所有數據倉庫都將被納入數據湖倉”。"}]},{"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","text":"11 月 12 日，Snowflake 做出迴應，發佈了自己的測試結果，同時稱 Databricks 公佈的性能比較結論缺乏完整性，而且研究本身也存在缺陷。Snowflake 公司創始人還強調這種基準測試沒什麼意義，在這個年代發佈數據庫基準測試結果是“將正常的技術交流變成了缺乏完整性的營銷噱頭”。"}]},{"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","text":"11 月 15 日，Databricks 的創始人再次在其公司博客上給予迴應，指責 Snowflake 爲了測試結果竟然改了 TPC-DS 的輸入數據，表示有些人不僅作弊還是“酸葡萄”。"}]},{"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","text":"這場戰鬥，雙方的企業創始人紛紛親自下場，可謂招招致命刀刀見血。大多數軟件供應商永遠不會滿足於第二名，這也意味着 Snowflake 和 Databricks 之間的激烈鬥爭可能纔剛剛開始。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"事情經過"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"第一回合：Databricks 出擊"}]},{"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","text":"Databricks 正着力推動一種名爲“數據湖倉”的新型架構，支持者稱這種架構甚至能夠消除對於數據倉庫的直接需求、顛覆幾十年來的行業標準，其意義堪比出現了一款能夠直接幹掉谷歌 Chrome 的新型瀏覽器設計方案。"}]},{"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","text":"信心滿滿的 Databricks 挑上的第一個對手，就是 Snowflake——只要能用自己爲雲時代重新設計的數據湖倉技術擊敗最強在位者，價值 1070 億美元的市場就將盡歸己有。"}]},{"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","text":"11 月 2 日，Databricks 宣佈經過事務處理性能委員會（簡稱 TPC）這家獨立行業組織的驗證確認，Databricks 的系統性能可達行業內最接近的其他數據倉庫競爭對手的 2.2 倍。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.geekbang.org\/wechat\/images\/b4\/b48bde6ae316a04bc989efa92ed830f7.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":false,"pastePass":false}},{"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","text":"Databricks 在博客中聲稱，在經典提取 - 轉換 - 加載（ETL）流程的過濾與處理方面，其智能湖倉方案取得了超越 Snowflake 數據倉庫方案的性能表現。此次比較採用了 TPC-DS 的基準測試並得到審計認證，其中 Databricks 實現了 3294 萬 1245 QphDS @ 100TB 的成績，打破了阿里巴巴定製系統此前保持的 1486 萬 1137 QphDS @ 100 TB 的世界紀錄。"}]},{"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","text":"Databricks 還宣佈巴塞羅那超級計算中心（BSC）的研究團隊運行了另外一項不同的基準性能比較，並發現 Databricks SQL（lake house）在同等規模下的速度可達 Snowflake 方案的 2.7 倍。研究團隊在 Databricks 基準測試中使用到兩種不同模式：按需與競價（即使用可靠性較低、但成本同樣較低的競價實例）。Databricks 在按需模式下的成本爲 Snowflake 的 1\/7.4，在競價模式下則可達到後者的 1\/12。"}]},{"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","text":"公司 CEO Ali Ghodsi 在採訪中表示，"},{"type":"text","marks":[{"type":"strong"}],"text":"“我們基本上已經成功證明在數據湖倉的技術對抗中擊敗了 Snowflake。”"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"第二回合：Snowflake 還手"}]},{"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","text":"Snowflake 公司創始人 Benoit Dageville 與 Thierry Cruanes 很快做出迴應，發表了一篇《行業標杆 誠信競爭》的博客文章。文中表示 Databricks“發佈的 Snowflake 結果不透明、未經審計且無法重現。而且，這些結果也與我們的內部基準測試結果和客戶體驗完全相悖。”"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.geekbang.org\/wechat\/images\/d6\/d6822b8b5021ce4b8f939ce3c4373e24.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":false,"pastePass":false}},{"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","text":"TPC-DS 基準測試會對體量爲 100 TB 的 TPC-DS 數據庫運行 99 次查詢。"}]},{"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","text":"Snowflake 對由巴塞羅那研究團隊測得的上述 Databricks-Barcelona 結果提出異議，並自行重現了測試內容："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.geekbang.org\/wechat\/images\/c1\/c1c084d08ed14b7917e7c4b4e2b24d38.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":false,"pastePass":false}},{"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","text":"“配置全部爲默認，所有查詢都在一套 4XL 數據倉庫上運行，總時長爲 3760 秒；連續運行兩輪，取最佳運行時間。可以看到，Snowflake 的實際結果達到 Databricks 報告結果的 2 倍多。而且這裏使用的還只是 4XL 數據倉庫，規模僅爲 Databricks 測試中所用倉庫的一半。”"}]},{"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","text":"雖然結果有所變化，但 Databricks 的性能領先地位並沒有動搖。不過 Snowflake 目前正在開發 5XL 倉庫技術，並宣稱“我們現階段的 5XL 倉庫在總運行時間上大大優於 Databricks（2597 秒對 3527 秒）。未來在推出通用版本時，各項水平還將進一步提升。”"}]},{"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","text":"Databricks 公司還強調，巴塞羅那研究團隊公佈的結果證明其產品性價比遠高於 Snowflake："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.geekbang.org\/wechat\/images\/13\/139e1ae710c6168021345fa1ae144594.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":false,"pastePass":false}},{"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","text":"Snowflake 這邊的兩位創始人當然也不認可 Databricks 的性價比結論，表示其中存在誤導性。“我們在 AWS-US-WEST 雲區域內運行的 4XL 倉庫標準版的按需模式價格爲每小時 256 美元。由於 Snowflake 產品按秒計費，所以運行整個基準測試只需要 267 美元，絕不是 Databricks 方面報告的 1791 美元。”"}]},{"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","text":"下圖所示爲 Databricks 宣稱成本與 Snowflake 實際成本比較："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.geekbang.org\/wechat\/images\/c9\/c94e191bcf774e9d3b3a4ca7e872afcc.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":false,"pastePass":false}},{"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","text":"所以跟性能的情況類似，"},{"type":"text","marks":[{"type":"strong"}],"text":"Databricks 的表現確實比 Snowflake 更好"},{"type":"text","text":"，但好不了那麼多。"}]},{"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","text":"而 Snowflake 創始人們認爲，“如果使用標準版定價，Snowflake 與 Databricks 在性價比方面就基本相當了：對於此次提交給性能委員會的基準測試，兩套方案同樣運行 3527 秒後的按需成本分別爲 267 美元與 275 美元。”"}]},{"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","text":"他們還鼓勵感興趣的朋友自己嘗試運行 Snowflake TPC-DS 基準測試，驗證到底是誰在信口雌黃。只需點擊幾下鼠標再等上一個小時左右，就能得出靠譜的結論。Snowflake 本身“不會發布綜合行業基準，因爲這些結果起不到任何有益客戶的作用。”"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"第三回合：Databricks 再次回噴"}]},{"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","text":"幾天後，Databricks 的創始人又親自下場撕 Snowflake，表示自己做的就是“客戶至上”的基準測試，並且認爲 Snowflake 準備的 TPC-DS 數據集有問題。Databricks 利用官方的數據集、同樣的硬件，發現測試的速度慢了一倍，和巴塞羅那研究團隊的測試速度差不多。"}]},{"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","text":"總之，“我們將官方 TPC-DS 數據集加載到 Snowflake 中，對運行功率測試所需的時間進行計時，結果比 Snowflake 在他們的博客中報告的時間長 1.9 倍”。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https:\/\/static001.geekbang.org\/wechat\/images\/6c\/6c7ce11971bc8def8e1782e817042e47.png","alt":null,"title":null,"style":null,"href":null,"fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"基準測試的意義有多大？"}]},{"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","text":"Snowflake 公司創始人在回擊中強調他們不願意參與這種“與現實體驗完全脫節、只爲打壓競爭對手而存在的基準測試之爭，這種行爲不符合我們客戶至上的核心價值觀。”"}]},{"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","text":"他們表示：“二十年前，數據庫行業裏每隔幾個月就會有一次破基準測試記錄的事情發生。這在當時是大家的首要任務… 但現在數據庫行業停止發佈新結果也是有原因的。”"}]},{"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","text":"“任何擁有一定從業經歷的朋友都有相同的體會，基準性能競賽只會分散企業爲客戶打造優質產品時的專注度。”再說回 Databricks 公佈的實例，“儘管 Databricks 的結果正由事務處理性能委員會（TPC）進行審計，但爲了在比較中佔據優勢，他們已經把正常的技術交流變成了缺乏完整性的營銷噱頭。”"}]},{"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","text":"卡耐基梅隆大學副教授 Andy Pavlo 對此也表示，“在企業層面，也許有些 CIO 會關心產品在性能委員會那邊的官方排名，但排名結果對實際銷售的影響並不大。”"}]},{"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","text":"雖然事務處理性能委員會的影響力近年來一直不斷下降，但在行業內的號召力仍然不容小覷。該委員會建立於上世紀八十年代，旨在爲數據庫性能評估提供中立的裁定指導。性能委員會會首先發布供各家數據庫廠商運行的系統基準，之後再對結果進行小組審議與官方認證。"}]},{"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","text":"但隨着整個行業的爆發式增長與競爭烈化，這些性能基準不但沒有降低人們的認知門檻、反而加劇了混亂與爭吵。例如，部分供應商開始大肆宣揚並未得到性能委員會正式認證的測試結果。Databricks 表示，他們發佈的最新結果已經得到性能委員會的“審計與公佈”。性能提升比例相當可觀，足以讓一部分仍在猶豫的潛在客戶下定決心。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"顛覆和變革是一場持久戰"}]},{"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","text":"關注大數據和數據倉庫領域動態的朋友可能有印象，本輪只是 Databricks 與 Snowflake 業務交鋒的又一個新回合。"}]},{"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","text":"這場衝突勢必會加劇數據倉庫和湖倉一體技術之間的戰鬥。"}]},{"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","text":"Databricks 最初主要是一家數據湖公司，但一直在添加數倉功能，最終走向湖倉一體。Snowflake 則是反過來的，作爲一家數倉起家的公司，卻一直忙於擁抱數據湖功能。湖倉一體作爲一個新興架構，很多企業目前還在早期探索階段。"}]},{"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","text":"雖然基準測試沒有二十年前那麼有影響力，但 Databricks 的測試結果值得關注。"}]},{"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","text":"“從長遠來看，所有數據倉庫都將被納入數據湖倉，”Databricks 的聯合創始人兼首席執行官 Ali Ghodsi 說。“這不會在一夜之間發生——這些東西會共存一段時間——但這個官方的世界紀錄清楚地證明，在價格和性能上，數據湖倉完勝數據倉庫。”"}]},{"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","text":"而"},{"type":"link","attrs":{"href":"https:\/\/mp.weixin.qq.com\/s?__biz=MjM5MDE0Mjc4MA==&mid=2651090633&idx=2&sn=9041d6d7b134230132adeeb098b8d386&scene=21#wechat_redirect","title":"","type":null},"content":[{"type":"text","text":"大數據平臺領域專家關濤在回覆 InfoQ 的採訪中"}]},{"type":"text","text":"也曾講道：“現在大多數企業都還沒有用到湖倉一體的新架構，他們要麼選擇了數據湖方案，要麼選擇了數倉方案。但大方向之下，業內對湖倉一體的整體大方向是高度達成一致的。"}]},{"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","text":"湖倉一體的興起本質上是由用戶訴求推動的，大家希望得到更好的數據治理和管理能力，同時又希望有更好的靈活性，特別是隨着 AI 的興起，完全純數倉的二維關係表已經無法承接半 \/ 非結構化數據的處理，AI 引擎不可能只跑在純數倉模型上。所以湖倉一體一定是未來的發展趨勢。做數倉的會有更多數據湖屬性，做數據湖的也會有更多的數倉屬性，最後根據實際需求去找到中間的平衡。”"}]},{"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","text":"另外，Databricks 已經從投資者手中籌得 35 億美元，專門用於聘請頂尖人才、打造競爭產品，可謂與 Snowflake 勢不兩立。Michalis Petropoulos 於今年 6 月加盟 Databricks 並出任高級工程總監。之前，他曾經領導過谷歌旗下的 BigQuery 團隊並監督 Amazon Redshift 項目。此外，曾在谷歌領導 Spanner 團隊的 Sridhar Machiraju 也在 11 月加入並擔任公司高級工程總監。"}]},{"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","text":"這還只是新陣容中很小的一部分，過去一年來已經有十幾名前亞馬遜、谷歌、Snowflake 以及 IBM 員工加入到 Databricks 陣營。後續預計還將有更爲龐大的招聘計劃：谷歌工程總監 Amit Shukla 將於本月晚些時候加入。"}]},{"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","text":"Databricks 公司聯合創始人 Reynold Xin 宣稱，“我們的核心數據倉庫團隊……在實際規模上可能已經超越了 Snowflake 那邊。”"}]},{"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","text":"在最近的幾輪融資、性能委員會的認可以及衆多新員工加入的利好加持之下，Databricks 的發展勢頭無疑頗爲強勁。截至 8 月 31 日，該公司年經常性收入已經超過 6 億美元，由此也能看出人們對 Databricks 的數據湖倉模型確實充滿期待。"}]},{"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","text":"但前路仍然艱難。雖然 Ghodsi 言之鑿鑿，認爲數據湖倉將給整個數據倉庫市場帶來顛覆性、甚至毀滅性的變革，但要想真的幹掉領域內的頭部廠商之一，並全盤取代他們長期受到歡迎的技術方案，單憑性能委員會的一份認證顯然還遠遠不夠。至少過去二十年來，無數企業級技術的迭起興衰已經反覆證明了這一點。"}]},{"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":"strong"}],"text":"延伸閱讀："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"link","attrs":{"href":"https:\/\/databricks.com\/blog\/2021\/11\/02\/databricks-sets-official-data-warehousing-performance-record.html","title":"","type":null},"content":[{"type":"text","text":"https:\/\/databricks.com\/blog\/2021\/11\/02\/databricks-sets-official-data-warehousing-performance-record.html"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"link","attrs":{"href":"https:\/\/www.snowflake.com\/blog\/industry-benchmarks-and-competing-with-integrity\/","title":"","type":null},"content":[{"type":"text","text":"https:\/\/www.snowflake.com\/blog\/industry-benchmarks-and-competing-with-integrity\/"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"link","attrs":{"href":"https:\/\/databricks.com\/blog\/2021\/11\/15\/snowflake-claims-similar-price-performance-to-databricks-but-not-so-fast.html","title":"","type":null},"content":[{"type":"text","text":"https:\/\/databricks.com\/blog\/2021\/11\/15\/snowflake-claims-similar-price-performance-to-databricks-but-not-so-fast.html"}]}]},{"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":"strong"}],"text":"InfoQ 採訪：現在是採用湖倉一體的好時機嗎？"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"link","attrs":{"href":"https:\/\/www.infoq.cn\/article\/pb09krdg9azagqh4ls4x","title":"","type":null},"content":[{"type":"text","text":"https:\/\/www.infoq.cn\/article\/pb09krdg9azagqh4ls4x"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}}]}