外文文獻(xiàn)翻譯---基于 web 的分析系統(tǒng)

1、<p><b>　　文獻(xiàn)翻譯</b></p><p>　　基于 Web 的分析系統(tǒng)</p><p>　院（系）名稱信息工程學(xué)院</p><p>　專(zhuān) 業(yè) 名 稱軟件工程</p><p><b>　　英文譯文</b></p><p>　　基于 Web 的分析系統(tǒng)<

2、;/p><p><b>　　馬克斯科特，約翰琳</b></p><p><b>　　1 摘要</b></p><p>　　在使用分析型數(shù)據(jù)庫(kù)時(shí)，分析人員將數(shù)據(jù)歸入公用組，并嘗試確定條件變化時(shí)產(chǎn)生的結(jié)果。例如，提高產(chǎn)品價(jià)格會(huì)增加單位利潤(rùn)，但可能會(huì)減少銷(xiāo)量????ù會(huì)產(chǎn)生較高還是較低的總利潤(rùn)？或者，聯(lián)邦貼現(xiàn)率的下降會(huì)如

3、何影響房地產(chǎn)貸款的收益？為了幫助分析人員根據(jù)歷史趨勢(shì)做出有根據(jù)的預(yù)測(cè)，Microsoft 在 SQL Server 2000 中提供了分析服務(wù)，在 SQL Server 7.0 中提供了 OLAP 服務(wù)。這些服務(wù)都提供 OLAP 功能，能夠?qū)⒋鎯?chǔ)在 SQL Server（或任何其他 OLE DB 兼容的數(shù)據(jù)源）上的數(shù)據(jù)處理成多維數(shù)據(jù)結(jié)構(gòu)，稱為多維數(shù)據(jù)集。多維數(shù)據(jù)集簡(jiǎn)化了趨勢(shì)分析和建立實(shí)體間交互方式聯(lián)系的過(guò)程。例如，房地產(chǎn)投資者采用現(xiàn)金流

4、模型來(lái)區(qū)分一組具有共同特征（如：地產(chǎn)類(lèi)型、地理位置和利率范圍）的貸款，并預(yù)測(cè)各種事件的影響。如果貸款提前償還或者借款人違約，后果將會(huì)如何？此類(lèi)不可預(yù)測(cè)的事件會(huì)如何影響貸款所擔(dān)保的債券的收益。</p><p>　　從包含幾百筆貸款的清單中選擇并區(qū)分具有分析特征的貸款是需要相當(dāng)技巧的。分析服務(wù)和 OLAP 服務(wù)有助于在各組貸款間建立聯(lián)系，以便分析人員能夠建立貸款假設(shè)模型。為了幫助客戶的房地產(chǎn)分析人員預(yù)測(cè)商業(yè)抵押證券的

5、業(yè)績(jī)，我們的開(kāi)發(fā)小組需要設(shè)計(jì)一個(gè)以各種方式（如：利率、到期期限或地產(chǎn)位置）來(lái)簡(jiǎn)化貸款分類(lèi)的系統(tǒng)。其界面應(yīng)易于學(xué)習(xí)和使用。而且，所開(kāi)發(fā)的系統(tǒng)需要在 Internet 上進(jìn)行安全的部署。為了滿足這些要求，開(kāi)發(fā)小組選擇了分析服務(wù)。</p><p>　　2 在Web上部署Office</p><p>　　在選定了后端技術(shù)后，開(kāi)發(fā)小組開(kāi)始制訂實(shí)現(xiàn)前端界面的計(jì)劃。多數(shù)金融分析人員使用 Microso

6、ft Excel，他們對(duì)其界面比較熟悉，感覺(jué)也很舒服。Excel 包括數(shù)據(jù)透視表服務(wù)，能夠允許分析人員連接到分析服務(wù)數(shù)據(jù)庫(kù)。Excel 的拖放界面提供了對(duì)多維數(shù)據(jù)的簡(jiǎn)單和直觀的訪問(wèn)，并不要求用戶進(jìn)行深入的培訓(xùn)。而且，通過(guò)使用 Excel 的制圖功能，用戶能夠以圖和表的形式表示數(shù)據(jù)。所以，對(duì)于前端界面，小組的首選是 Microsoft Office XP 中的 Excel 2002。Excel 數(shù)據(jù)透視表服務(wù)瀏覽一個(gè)分析服務(wù) OLAP 多

7、維數(shù)據(jù)集的情形。</p><p>　　2.1 使用OWC在Web上部署Office</p><p>　　如果所有的客戶端用戶在同一幢大樓內(nèi)一起工作，并通過(guò)同一個(gè)局域網(wǎng)訪問(wèn)分析服務(wù)器，Excel 會(huì)是不錯(cuò)的選擇。但用戶需要和辦公地點(diǎn)散布于世界各地的不同組織共享應(yīng)用程序，因此開(kāi)發(fā)小組需要一個(gè)用戶可以通過(guò) Internet 訪問(wèn)且類(lèi)似于 Excel 的組件。該小組發(fā)現(xiàn) Office Web 組

8、件 (OWC) 能夠滿足這一需要。OWC 是一組能夠在 Web 頁(yè)上使用并提供 Office 功能的 ActiveX 控件。OWC 數(shù)據(jù)透視表組件是 Excel 中數(shù)據(jù)透視表服務(wù)的 Web 版本；數(shù)據(jù)透視表使用數(shù)據(jù)透視表服務(wù)，并要求在運(yùn)行前安裝數(shù)據(jù)透視表服務(wù)。但沒(méi)有 Excel，OWC 數(shù)據(jù)透視表也能工作。</p><p>　　數(shù)據(jù)透視表能夠從分析服務(wù)器上檢索多維數(shù)據(jù)并將這些數(shù)據(jù)顯示在一個(gè)交互的拖放界面上。已安裝

9、 Microsoft Internet Explorer (IE) 4.01 或以上版本的用戶可以使用 OWC 對(duì)分析服務(wù)數(shù)據(jù)進(jìn)行分析，而不必安裝額外的組件軟件。圖 2顯示了外觀和操作都類(lèi)似于熟悉的 Excel 界面的 OWC 數(shù)據(jù)透視表客戶端界面。OWC 數(shù)據(jù)透視表也提供了智能緩存，通過(guò)減少數(shù)據(jù)透視表從網(wǎng)絡(luò)到服務(wù)器的往返行程次數(shù)而提高性能。所以，通過(guò)有效使用分析服務(wù)，數(shù)據(jù)透視表能夠減少數(shù)據(jù)傳輸并提高效率。</p><

10、;p>　　雖然 OWC 提供了我們開(kāi)發(fā)小組的項(xiàng)目需要的全部東西，但當(dāng)我們?cè)噲D在 Internet 上部署 OWC 時(shí)，我們遇到了難題。首先是 OWC 的運(yùn)行平臺(tái)問(wèn)題。Office XP 版的 OWC 要求使用 Microsoft Data Access Components (MDAC) 2.6 或以上版本。而許多服務(wù)訂戶使用 Windows NT Workstation 4.0 作為其操作系統(tǒng)，如果要安裝 MDAC 2.6，還必

11、須安裝 Service Pack 6 (SP6)。使用 OWC 的一個(gè)主要吸引力在于我們認(rèn)為它能夠?qū)崿F(xiàn)無(wú)縫的部署。我們發(fā)現(xiàn)雖然能夠自動(dòng)處理安裝 Service Pack，但該過(guò)程需要重新啟動(dòng)，非常麻煩。以后，Microsoft 提供了一個(gè)使用 SP4 的 OWC 組件修訂版本，但我們同時(shí)也在開(kāi)發(fā)自己的應(yīng)用程序，在金融機(jī)構(gòu)嚴(yán)格控制的客戶端網(wǎng)絡(luò)上部署 Service Pack 是一個(gè)很大的困難。因此，需要在操作系統(tǒng)上使用特定 Service

12、 Pack 的解決方案是不可行的。</p><p>　　其次，我們小組遇到了連接問(wèn)題。OWC 要求直接連接分析服務(wù)數(shù)據(jù)源。OWC 使用默認(rèn)的 2725 端口直接和分析服務(wù)器通信，對(duì)于使用防火墻的機(jī)構(gòu)來(lái)說(shuō)，這是個(gè)問(wèn)題。先，我們?cè)噲D使用 HTTP 連接和服務(wù)器通過(guò) 80 端口進(jìn)行連接以解決連接問(wèn)題。該連接通過(guò) Web 瀏覽器使用的同一個(gè)端口來(lái)提供訪問(wèn)。對(duì)于額外的安全性，分析服務(wù)還能夠使用安全套接字層 (SSL)，通過(guò)

13、 443 端口進(jìn)行連接。大多數(shù)組織同時(shí)打開(kāi) 80 端口和 443 端口以便用戶訪問(wèn) Internet。（有關(guān)使用 HTTP 的更多信息，參見(jiàn)位于 http://support.microsoft.com/default.aspx?scid=kb;en-us;q279489 的 Microsoft 文章 "INF: How to Connect to Analysis Service 2000 By Using HTTP Con

14、ection"。）說(shuō)明了使用 OWC 連接到分析服務(wù)器涉及的問(wèn)題。</p><p>　　然而，HTTP 連接的執(zhí)行產(chǎn)生了一些難以克服的困難。我們測(cè)試顯示，通過(guò) 80 端口進(jìn)行連接要明顯慢于直接連接。因?yàn)槎嗑S數(shù)據(jù)集需要向客戶端提供大量的數(shù)據(jù)，性能的降低使 OWC 的使用很不現(xiàn)實(shí)。</p><p>　　2.2 尋找替代方案</p><p>　　接下來(lái)，我們的

15、小組考慮使用 ADO-MD 和 MDX 查詢創(chuàng)建自定義界面。您可使用 OPENROWSET 命令直接查詢分析服務(wù)多維數(shù)據(jù)集。（有關(guān)查詢分析服務(wù)的信息，參見(jiàn)位于 http://support.microsoft.com/default.aspx?scid=kb;en-us;q218592 的 Microsoft 文章 "HOWTO:SQL Server 7 Distributed Query with OLAP Server&q

16、uot;。）OPENROWSET 允許您從包括分析服務(wù)在內(nèi)的任何 OLE DB 源上查詢數(shù)據(jù)。這種靈活性能夠讓我們使用 ADO 查詢分析服務(wù)。分析服務(wù)使用的 OLE DB 提供程序 MSOLAP 將多維數(shù)據(jù)轉(zhuǎn)換成 ADO 能夠用來(lái)同前端的應(yīng)用程序進(jìn)行數(shù)據(jù)通信的標(biāo)準(zhǔn)行集。這種自定義解決方案的問(wèn)題在于創(chuàng)建具有 OWC 和 Excel 外觀的直觀而且互動(dòng)的界面是一件非常復(fù)雜的工作。雖然開(kāi)發(fā)小組能夠創(chuàng)建這樣的界面，但所花時(shí)間長(zhǎng)、費(fèi)用高，且需要不

17、斷進(jìn)行維護(hù)，因此該解決方案不具備可行性。</p><p>　　開(kāi)發(fā)小組也研究了幾個(gè)第三方的解決方案。很多第三方解決方案是幫助用戶生成一個(gè)查詢，然后執(zhí)行它來(lái)查看結(jié)果。這種方式雖然有效地利用了系統(tǒng)資源，但達(dá)不到 Excel 和 OWC 的拖放界面同樣的交互式效果。所以，雖然這些解決方案各具優(yōu)勢(shì)，但沒(méi)有一個(gè)能夠完全滿足本項(xiàng)目的要求。在我們小組將自定義開(kāi)發(fā)成本加入第三方軟件的總成本考慮時(shí)，我們決定重新尋找替代方案。<

18、;/p><p>　　2.3 使用 Web 瘦客戶端訪問(wèn)多維數(shù)據(jù)</p><p>　　開(kāi)發(fā)小組成員最后選用 Microsoft SQL Server Resource Kit，為我們的難題找到了解決方案：分析服務(wù) Web 瘦客戶端瀏覽器。（要在資源工具箱 CD-ROM 上訪問(wèn) Web 瘦客戶端，參見(jiàn)資源工具箱第 39 章的參考信息，該信息位于 http://www.microsoft.com/

19、technet/treeview/default.asp?url=/technet/prodtechnol/sql/reskit/sql2000/part11/c3961.asp。）Web 瘦客戶端使用 Active Server Page (ASP) 連接到分析服務(wù)器、將多維數(shù)據(jù)轉(zhuǎn)換成 HTML，并將數(shù)據(jù)傳遞給客戶端。說(shuō)明了 Web 瘦客戶端顯示貸款數(shù)據(jù)子集的情況。</p><p>　　Web 瘦客戶端需要 I

20、E 5.0 或以上版本。因?yàn)榭蛻舳瞬恢苯舆B接到分析服務(wù)計(jì)算機(jī)，所以客戶端不需要 MDAC 2.6。由于多數(shù)客戶端訂戶具有 IE 5.0，所以我們不必在訂戶的個(gè)人計(jì)算機(jī)上部署 Service Pack。圖 5顯示了 Web 瘦客戶端使用的結(jié)構(gòu)。Web 瘦客戶端使用 ASP 從 Microsoft IIS 服務(wù)器 — 而不是客戶端 — 來(lái)查詢多維數(shù)據(jù)集。Web 瘦客戶端帶有 ASP 頁(yè)，使得您能夠在 IIS 服務(wù)器上部署。因?yàn)橹挥羞\(yùn)行 AS

21、P 的 IIS 服務(wù)器才能夠連接到分析服務(wù)數(shù)據(jù)庫(kù)，所以，您能夠使用一個(gè)連接來(lái)保證通過(guò)防火墻進(jìn)行的多維數(shù)據(jù)集訪問(wèn)，并且可以將該連接局限于 Web 服務(wù)器和數(shù)據(jù)庫(kù)服務(wù)器。這種安排創(chuàng)建了一個(gè)高效和易于保護(hù)的連接。對(duì)客戶端的唯一連接是標(biāo)準(zhǔn)的 HTML 連接，這能夠減少防火墻的影響。</p><p>　　Web 瘦客戶端將查詢的數(shù)據(jù)放到一個(gè)網(wǎng)格狀的 HTML 表結(jié)構(gòu)中，并將數(shù)據(jù)發(fā)送到瀏覽器。用戶通過(guò) JavaScript

22、和數(shù)據(jù)進(jìn)行交互。通過(guò)使用透明層，用戶可以將維拖到多維數(shù)據(jù)集中進(jìn)行數(shù)據(jù)操作、通過(guò)維向下追溯并顯示和分析其需要的數(shù)據(jù)。</p><p>　　如多數(shù)設(shè)計(jì)決策那樣，使用 Web 瘦客戶端需要做出某些折衷。因?yàn)樵?Web 服務(wù)器域和后端域之間不存在信任關(guān)系。您不能在系統(tǒng)中擴(kuò)展 Active Directory (AD) 作為其驗(yàn)證服務(wù)。這樣的信任關(guān)系會(huì)給入侵者提供一種破壞后端域安全的方法。因此，系統(tǒng)必須建立一個(gè)匿名用戶來(lái)訪

23、問(wèn)分析服務(wù)器上的數(shù)據(jù)。如果對(duì)數(shù)據(jù)的訪問(wèn)取決于用戶，則不能使用基于分析服務(wù)的角色安全性來(lái)控制對(duì)數(shù)據(jù)的訪問(wèn)。然而，如果系統(tǒng)允許所有用戶通過(guò)一個(gè)安全上下文進(jìn)行數(shù)據(jù)訪問(wèn)，則使用 Web 瘦客戶端是安全的，而且容易部署。</p><p>　　Web 瘦客戶端也有其他缺點(diǎn)。OWC 通過(guò)使用數(shù)據(jù)透視表的智能緩存把從 Web 瀏覽器到 OLAP Web 服務(wù)器的往返行程數(shù)減少到最小來(lái)獲取更多的數(shù)據(jù)。但 Web 瘦客戶端是從服務(wù)器

24、端提取數(shù)據(jù)的，因?yàn)閿?shù)據(jù)在 Web 瀏覽器中并不緩存，所以每次對(duì)數(shù)據(jù)的更改都要求瀏覽器向 Web 服務(wù)器提出新的請(qǐng)求。當(dāng)處理的數(shù)據(jù)量很大時(shí)，這個(gè)進(jìn)程會(huì)很緩慢。OWC 還具有豐富的對(duì)象模型，您可以對(duì)之編寫(xiě)自定義代碼。而 Web 瘦客戶端使用 JavaScript，很難自定義。因?yàn)?Web 瘦客戶端是資源工具箱內(nèi)包含的“自由代碼”，所以 Microsoft 并不像支持 Excel 或 OWC 那樣對(duì) Web 瘦客戶端提供相同程度的支持。而且，

25、Web 瘦客戶端要求客戶端腳本，如果 Web 瀏覽器已經(jīng)更新或更改，它會(huì)產(chǎn)生錯(cuò)誤。此外，Web 瘦客戶端不直接支持對(duì)服務(wù)器的回寫(xiě)。幸好，我們小組的應(yīng)用程序不要求回寫(xiě)， 所以，這個(gè)缺失的功能并不妨礙我們使用 Web 瘦客戶端。在對(duì)其優(yōu)點(diǎn)和缺點(diǎn)進(jìn)行評(píng)估后，我們決定部署 Web 瘦客戶端。這個(gè)解決方案在生產(chǎn)環(huán)境中使用效果不錯(cuò)，并得到了用戶的廣泛認(rèn)可。</p><p>　　3 為任務(wù)選擇合適的客戶端</p>

26、<p>　　我們的開(kāi)發(fā)小組在本項(xiàng)目中得到的經(jīng)驗(yàn)教訓(xùn)能夠?yàn)樵谄髽I(yè)中部署 OLAP 解決方案提供有價(jià)值的指導(dǎo)。Excel 提供了功能強(qiáng)大而且熟悉的客戶端，您可以在局域網(wǎng)中進(jìn)行部署，但需要實(shí)時(shí)連接 OLAP 服務(wù)器。在 Intranet 內(nèi)部署分析服務(wù)客戶端時(shí)，OWC 的使用效果不錯(cuò)，因?yàn)槟梢暂p松控制客戶端平臺(tái)，并且能夠在 Intranet 內(nèi)安全地打開(kāi)端口。當(dāng)存在防火墻，并且您希望盡可能減少對(duì)用戶操作系統(tǒng)的影響時(shí)，Web 瘦

27、客戶端是一個(gè)很好的 Internet 解決方案。對(duì)于任何開(kāi)發(fā)項(xiàng)目，需要理解所開(kāi)發(fā)產(chǎn)品的使用者的業(yè)務(wù)要求和需要。通過(guò)概括要求并權(quán)衡所有選擇，可以發(fā)現(xiàn)能夠滿足客戶要求的正確解決方案。</p><p>　　摘自：馬克斯科特，約翰琳.基于 Web 的分析系統(tǒng)[A].微軟官方網(wǎng)站,2007.2</p><p><b>　　附：英文原文</b></p><p&

28、gt;　　Overview of Jscript8.0 Arrays</p><p>　　James street </p><p>　　1 Summary</p><p>　　When working with analytical databases, analysts organize data into common groups and try to d

29、etermine what would happen if things were different.</p><p>　　For example, would increasing a product's price?which would increase profit per unit but probably reduce number of units sold?yield a higher

30、or lower overall profit? Or how would a drop in the federal discount rate affect the yield of real estate loans? To help analysts make educated projections based on historical trends, Microsoft provides Analysis Services

31、 in SQL Server 2000 and OLAP Services in SQL Server 7.0. These services provide OLAP capability and can process data stored in SQL Server (</p><p>　　For example, real estate investors use cash-flow modeling

32、to isolate a group of loans that have common characteristics (e.g., types of properties, geographic area, range of interest rates) and project the effects of different kinds of events. What will happen if loans mature mo

33、re rapidly than expected or if the borrowers default? And how might such unpredictable events affect the yield of bonds that the loans secure?</p><p>　　Selecting from lists that can include hundreds of loans

34、 and isolating the loans that have the characteristic that you're analyzing can be tricky. Analysis Services and OLAP Services can help correlate these groups of loans so that analysts can model loan assumptions. To

35、help a client's real estate analysts project the performance of commercial mortgage-backed securities, our development team needed to devise a system that simplified the grouping of loans in different ways?such as by

36、 their intere</p><p>　　2　Deploy Office OWC</p><p>　　Having settled on a back-end technology, the development team began working on a plan for implementing the front-end interface. Most financial

37、 analysts use Microsoft Excel and are familiar and comfortable with its interface. Excel includes PivotTable Service, which lets analysts connect to Analysis Services databases. Excel's drag-and-drop interface provid

38、es simple, intuitive access to multidimensional data without requiring users to have extensive training. And by using Excel's graphing capabil</p><p>　　2.1　Deploy Office OWC</p><p>　　Excel w

39、ould have been a fine choice?if all the client's users worked together in the same building and could access the Analysis server through the same LAN. But because the users needed to share the application from a vari

40、ety of organizations whose offices are scattered around the world, the team needed a component similar to Excel that users could access through the Internet. The team found the solution to this challenge in Office Web Co

41、mponents. OWC is a set of ActiveX controls that you can u</p><p>　　PivotTable can retrieve multidimensional data from an Analysis server and present the data in an interactive, drag-and-drop interface. Users

42、 who have Microsoft Internet Explorer (IE) 4.01 or later can use OWC to analyze Analysis Services data without installing additional component software. Figure 2 shows the OWC PivotTable client interface, which looks and

43、 works like the familiar Excel interface. The OWC PivotTable also provides intelligent caching, which improves performance by reducing the n</p><p>　　Although OWC provided everything our development team'

44、;s project needed, we encountered problems when we tried to deploy OWC across the Internet. The first problem was the platform that OWC runs on. The Office XP version of OWC requires Microsoft Data Access Components (MDA

45、C) 2.6 or later. Many of the service subscribers use Windows NT Workstation 4.0 as their OS, and to install MDAC 2.6, they also had to install Service Pack 6 (SP6). One of the primary attractions of using OWC was that we

46、 though</p><p>　　The second problem that our team encountered was connectivity. OWC requires a direct connection to the Analysis Services data source. OWC communicates directly with the Analysis server throu

47、gh the default port, 2725, which is a problem for organizations with firewalls. (For details about connecting to Analysis Services, see the Microsoft article "INF: TCP Ports Used by OLAP Services when Connecting Thr

48、ough a Firewall" at http://support.microsoft.com/default.aspx?scid=kb;en-us;q301901.)</p><p>　　First, we tried to solve the connectivity problem by using an HTTP connection to connect to the server thro

49、ugh port 80. This connection provides access through the same port that the Web browser uses. For added security, Analysis Services can alternatively connect through port 443 by using Secure Sockets Layer (SSL). Most org

50、anizations open both port 80 and port 443 to let users access the Internet. (For more information about using HTTP, see the Microsoft article "INF: How to Connect to Analysi</p><p>　　2.2　Looking for Alt

51、ernatives</p><p>　　Next, our team considered using ADO-MD and MDX queries to create a custom interface. You can query the Analysis Services cube directly through the OPENROWSET command. (For information abou

52、t querying Analysis Services, see the Microsoft article "HOWTO: SQL Server 7 Distributed Query with OLAP Server" at http://support.microsoft.com/default.aspx?scid=kb;en-us;q218592.) OPENROWSET lets you query da

53、ta from any OLE DB source, including Analysis Services. This flexibility would let us use ADO to query</p><p>　　The team also investigated several third-party solutions. Many third-party solutions help users

54、 build a query and execute it to see the results. Such an approach uses system resources efficiently but isn't as interactive as the Excel and OWC drag-and-drop interfaces. So although these solutions offered unique

55、advantages, none met the exact requirements of the project. When our team added custom development costs to the overall cost of the third-party software, we decided to look elsewhere.</p><p>　　2.3　Using the

56、Thin Web Client to Access Multidimensional Data</p><p>　　The development team members finally went to the Microsoft SQL Server Resource Kit and found the solution to our dilemma: the Analysis Services Thin W

57、eb Client Browser. (To access the Thin Web Client on the resource kit CD-ROM, see the reference in chapter 39 of The Thin Web Client uses Active Server Pages (ASP) to connect to the Analysis server, converts the multidi

58、mensional data to HTML, and passes the data to the client. Figure 4 shows the Thin Web Client displaying a subset of loan data.</p><p>　　The Thin Web Client requires IE 5.0 or later. Because the client doesn

59、't connect directly to the Analysis Services computer, the client doesn't need MDAC 2.6. Most of our client's subscribers have IE 5.0, so we didn't have to deploy service packs to the subscribers' PCs

60、. Figure 5 shows the architecture that the Thin Web Client uses. The Thin Web Client uses ASP to query the cube from the Microsoft IIS server rather than from the client. The Thin Web Client comes with ASP pages, which y

61、ou can dep</p><p>　　The Thin Web Client puts the queried data in a gridlike HTML table structure and sends the data to the browser. The user interacts with the data through JavaScript. By using transparent l

62、ayers, users can drag dimensions into the cube to manipulate the data, drill down through dimensions, and display and analyze the data they need.</p><p>　　Using the Thin Web Client requires some compromises,

63、 as most design decisions do. Because a trust relationship can't exist between the Web server domain and the back-end domain, you can't extend Active Directory (AD) as your authentication service through the syst

64、em. Such a trust relationship could provide a means for an intruder to violate security on the back-end domain. So the system must set up one anonymous user to access the data on the Analysis server. If access to the dat

65、a depends on the</p><p>　　The Thin Web Client also has other drawbacks. OWC uses PivotTable's intelligent caching to minimize trips from the Web browser to the OLAP Web server to get more data. But the T

66、hin Web Client draws data from the server side, and because data isn't cached at the Web browser, each change to the data requires the browser to make a new request from the Web server. When you're working with l

67、arge amounts of data, this process can be sluggish. Also, OWC has a rich object model against which you can pro</p><p>　　2.4　Using the Thin Web Client to Access Multidimensional Data</p><p>　　Th

68、e development team members finally went to the Microsoft SQL Server Resource Kit and found the solution to our dilemma: the Analysis Services Thin Web Client Browser. (To access the Thin Web Client on the resource kit CD

69、-ROM, see the reference in chapter 39 of The Thin Web Client uses Active Server Pages (ASP) to connect to the Analysis server, converts the multidimensional data to HTML, and passes the data to the client. Figure 4 shows

70、 the Thin Web Client displaying a subset of loan data.</p><p>　　The Thin Web Client requires IE 5.0 or later. Because the client doesn't connect directly to the Analysis Services computer, the client doe

71、sn't need MDAC 2.6. Most of our client's subscribers have IE 5.0, so we didn't have to deploy service packs to the subscribers' PCs. Figure 5 shows the architecture that the Thin Web Client uses. The Thin

72、 Web Client uses ASP to query the cube from the Microsoft IIS server rather than from the client. The Thin Web Client comes with ASP pages, which you can dep</p><p>　　The Thin Web Client puts the queried dat

73、a in a gridlike HTML table structure and sends the data to the browser. The user interacts with the data through JavaScript. By using transparent layers, users can drag dimensions into the cube to manipulate the data, dr

74、ill down through dimensions, and display and analyze the data they need.</p><p>　　Using the Thin Web Client requires some compromises, as most design decisions do. Because a trust relationship can't exis

75、t between the Web server domain and the back-end domain, you can't extend Active Directory (AD) as your authentication service through the system. Such a trust relationship could provide a means for an intruder to vi

76、olate security on the back-end domain. So the system must set up one anonymous user to access the data on the Analysis server. If access to the data depends on the</p><p>　　The Thin Web Client also has other

77、 drawbacks. OWC uses PivotTable's intelligent caching to minimize trips from the Web browser to the OLAP Web server to get more data. But the Thin Web Client draws data from the server side, and because data isn'

78、t cached at the Web browser, each change to the data requires the browser to make a new request from the Web server. When you're working with large amounts of data, this process can be sluggish. Also, OWC has a rich

79、object model against which you can pro</p><p>　　3　Choosing the Right Client for the Task</p><p>　　The lessons our development team learned from this project can provide a valuable guide for depl

80、oying OLAP solutions in an enterprise. Excel provides a capable, familiar client that you can deploy in a LAN but requires realtime connectivity to the OLAP server. OWC works well for deploying an Analysis Services clien

81、t in an intranet because you can easily control the client platform and open ports securely in an intranet. The Thin Web Client provides a good Internet solution when firewalls are in </p><p>　　From:Mark Sco