200947322 丸、發明說明： 【發明所屬之技術領域】 本發明係關於一種企業資訊系統整合服務分派與流控方法，特別為一 種應用在企業資訊服務整合平台架構中，提供各後端應用系統動態監測功 能與服務分派及流控方法功能。 【先前技術】 大型企業為了解決不同系統介接與資料分散的問題，便衍生出企業應 用軟體整合平台(Enterprise Application Integration)或以服務為導向之架構平 台(Service Oriented Architecture)之需求，企業應用軟體整合主要目的在於， 、©將企業内部各應用程式透過即時訊息整合之方式(Enterprise serviee Bus)， - 達到資訊透通性，企業便可即時且有效的運用這些資料，在内部行政與決 滚及各種商業業務競的場合。為提昇市場優勢，配合新業務的行銷及電 信業務推廣能達到訊息即時化，資訊整合化與流程自動化，在既有的資源 中創造更大的獲利能力，仰賴快速内部資源的整合及快速的客戶資料與服 務的整合與提供。 然而，現有的企業應用軟體整合的技術，雖然可在正常的情況下提供 即時訊息整合的功能，但是，卻未能完整的提供在例外的情況下提供錯誤 φ防止或自動修流量控管之能力，當某部份連結到整合平台上的後端系統發 生問題’而前端系統使用者(Call Center，Intemet Web Site,IVR)無法久等，而 _ 不停的發出請求，若整合平台沒有提供一錯誤偵測與流量控管之能力，則 系統終將因資源耗盡而全面產生問題。現今系統分析問題大多僅止於線上 - 記錄(l〇g)的功能，無法即時(realtime)判斷資訊流的異常現象，而在系統内部 即時做出決策與控管，無需系統維護人員介入。例如企業應用軟體整合可 提供使用者身份辨識，以過濾非法使用者，但是卻未能防止合法使用者惡 意或無心的破壞，如使用者不在預期下持續不斷地發出訊息，使得系統的 速度變慢’或資源被佔住，無法提供其他合法且正常的使用者使用。又如 某後端應用系統不正常工作，致使資訊流滯留在企業應用軟體整合系統 5 200947322 中，致使拖累整個整合平台系統導致無 費大量的咖與人力，事後找找出_的運^系歸理者往往要花 且讓系統管理者疲於奔命。因此，為使企業廊用大的雉護成本。 工作，須要有-套，能鱗時追树上各應科面^祕能有效率的 台不因各_統發生狀況而影響其它系統運作: 系統運作更觀，發揮最大效能。 技m應用軟體整合 失，由實3善=^中分析各純梅上運行㈣，树諸多缺 天貫非一良善之設計，而亟待加以改良。 新，ίίΓΓί麟上翻財搞魅的各魏點，g思加以改良創 合服務分_ π力研發谢件，企業資訊整 【發明内容】 本=月的目的在提供一種企業級軟體整合服務平台的自動管控，透過系 =線土自動_歡’並記錄及運麟树上介面制運作狀況，當線上 應用介通務發生_時，除—般告料，綠並會自動進人控管狀態， 不至於因單ϋ或少數介面出現問題而拖轉個服務平台。 鲁 本發月的第—目的在提供整合平台線上流量自動分派管理功能，自動分 派刖端應用系統加諸於本平台之負載。 本發明的第三目的在於將;^同純所提供之_介錢過此—平台整 理與轉換後能夠將不-致之通訊規約與資料格式，轉換為一致之規約與資 料格式’如此前端應用系統不再因為與後端系關之連接模式不同而耗費 研發人力與時程。 本發明的第四目的係透過四個層次(基層、轉換層、整合層與入口層)之 軟體架構’將各別系統所提供之介面(基層)由不同通訊規約(Protocd)與資料 内容轉換’成為中間層並可透過組裝將現有介面組成更大的介面資料(整合 層）’最後，再次轉換為提供服^務之入口層，提供單一資料介面格式與Pr〇t〇c〇l 200947322 給前端應用系統使用。 雜^雜獅狀輕纽整合服務分派錢㈣統及其方法，係 提供-種企業自動化分派與流控f理之整合服務平台，包括不同規約 (―)之人π層歡㈣s〇Ap，Hm厕，s〇cket )，各種不同規約 之入口賴組都可將前端請求分派至所有後端系統所提供之服務，形成一 個企業級虛擬資料庫’較複雜之功能可能需要整合一個以上之介面而提供 更多70服務’系統管理者使用劉覽器進行流控的參數設定。 本發明亦包含整合平台系統即時流控模組，其針對執行整合平台之應 用服務進行監測與流量控制。 ‘ '© 本發明亦提供應用於企業内，各應用系統間，-個完整且統合之資訊. _交換平台架構’達到統—控管與服務介面再之目心舉凡資料庫系統、 主機系統(VAX ’ IBM主機，HP主機)及主從式架構等皆可經由此整合平台 系統’提供整合性服務。前端系統經由呼叫本發明的入口層分派模組，此 模組兼具有收集介面執行狀態之功能，分派前端請求至整合層模組功能， -而整合層模組根據該請求内容判斷是否整合一個以上之子服務，而同時對 多個轉換層發出請求，各轉換層模組收到整合層模組之請求後，再轉換規 約_〇甽與資料格式向基賴出請求，這裡的基層指的直接連結後端所 φ提供之原始介面模組，如電信公司之市話受理綠介面、市嫌務系統介 面、行動業務帳務系統介面…等等。每一個請求都將做記錄，並將查詢結 果狀態即時儲存於資料財，並且每讀求成賴失敗狀_會回傳至入 口層與前端應用系統，以利識別並做進一步做流控處理。 本發明亦提供自祕分派架構麟組’主要在於管理前端顧系統經 由本發明的服務整合平台來存取後端系統，透過入口層，每一前端之請求結 果與狀態都會經由多層次分派模組處理並由入口層判斷後，交由管理模組 s己錄處理狀況，當任一服務功能發生問題時將該服務即會透過入口層進入 流控管理。 【實施方式】 200947322 請參閱圖-所示，麟本魏之企業資婦合赚分派㈣控系統及 其方法之架構圖，該架構包括： 一入口層101 ’該入口層101係提供單一規約與义^資料格式介面入 口 ’並於入口層101建置多層次分派模組及即時流控模組； -整合層1〇2’該整合層1〇2係整合不同子服務成為大的服務，並於 整合層102建置整合模組； 一轉換層1〇3 ’該轉換層103係將不同規約與資料格式轉換為統一格 式’並於轉換層103建置轉換模組； 一基層104，該基層104係應用與後端系統一之規約介接後端系統； © 一管理模組105，該管理模組105係設定系統參數。 - 透過上述之架構，即可提供以下功能： (1) 可避免因同步應用所產生的死鎖(DeadLock)間題； (2) 流量管理功能； (3) 自動化分派功能。 - 達成以上各功能之方法，說明於下： (1) 可避免因同步應用所產生的死鎖(Dead Lock)問題： 企業服務整合訊息平台提供不同介面之介接技術，藉由不同規約 ❹（Protocol)間資枓之轉換’達成系統間整合的目的，將不同服務介面整合後， 可在企魏雜供錢程巾提織_肛具，使縣始㈣碼(s_e Code)在系統中的商業邏輯得以抽離，達成端點對端點處理(㈣於End Pro⑽)功能’以大型企業來考量賴步的訊息匯流排顧鋪，可避免因 同步應用所產生的關(Dead 1磁)騎。企f服務整合平自提供四個單一 服務介面人d層(HTTP, SGAP，〗MS，SQCKET)，麵各單-介®即可存取後 端已介接系統之所有功能。 (2) 流量管理功能： 透過服務介面狀態朗(計算整合平台上各應用介面執伽應時間與 回應狀雜）’各顧介面可得到___之回應值，此值可為前日平均回應 200947322 時間或月平均值，這些值可透過計算資料庫中之L〇g資料得到，當然因每 個服務介面在各後端系統中所負貴的運作内容不一樣(如有些是單純查詢資 料庫，有些可能是異動，不同的軟硬體效能也不一樣），因此，每個介面所計算 出之運行回應平均時間值都不太一樣。 所謂流量管理是針對運行回應平均時間(Mn秒，Μη表示第η個介面之 平均回應時間值)所設定之某一服務介面所能接受之期待值f(Mn#)、，此期 待值即為系統可容忍該服務介面最長之回應時間。當某一服務介面回應時 間超過f(Mn)B夺，入口介面層便會記錄該值(c(n)=c(n)+l c⑻為第n個服務 介面連續超過該監測值f(Mn)之次數，此處是應用Hash table方式動態記錄)。 ❹ 而入口層101模組讀取透過管理者介面所設定之流控門檻值·τ(η)進行 -/;IL控(當T(n) > c(n)時），該服務介面立即進入流控，此時前端系統對這個服 務"面之所有請求都將被入口層1〇1模組拒絕，以降低整合平台之異常負 載。該服務介面進入了流控後，如何在後端恢復正常運作後，可自動恢復 該服務介面之運作是相當重要的，一般的設計是啟動另一測試模組針對該 介面定期做循環職，但這會有—問題，如果是異_服務介面就很難處 理(因只要一測試就會將資料寫入線上真實的資料庫中或進入一個複雜的異 動流程中’如申請一個ADSL門號，在-個龐雜的組織中是不可能為每個 φ異動的服務介面在線上主機中實作測試介面），因此，在此整合平台中設計 一時間區間流控模組，以解決此一問題。一個大型企業往往前端有數百甚 至數千個者’整合平台也可能同時間有上千個應用服務介面在運行， 每個應贿務魏瞬間都可脉飾當大的贿量，當顧務功能發生問 題時，如回應時間报長甚至不回應(鎖死），瞬間數百個甚至上千個請求全部 鎖死在整合平台上，如果沒有一個好的機制來自動解決此問題，不只是前 用者系統長時間等待回應結果，整合平台的資源(cpu，Mem〇ry，I/〇)等都 可能會因少數服務功能不正常運作而耗盡軟硬體資源，使得其它正常之服 務;1面功也無法對前端祕做服務，因此除了細-般之執行緒(Thread) 群組控制系統資源使用外並適時發出告警通知已發生障礙之目標系統外(透 200947322 過email，簡訊.·等）’需更進一步應用時間區間流控模組加以管控。 請參閱圖二所示，係本發明表示某一服務介面功能c⑻>τ⑻，該功能 已進入時間區間流控，τ值由管理介面設定可為任意值，但通常與後端系統 修復及介面時間有關，一般而言，以分為單位，如5秒在，，測試⑽，抓時 •k’絕大射明端請求到達入口層1〇1後即馬上回應前端訊息:，，此服務介面 異常無法服務請稍後再試”，此時，尸、有管理介面設定之某一請求數l可通 過入口層101到達後端系統，因此，除了流控外更肩負測試後端是否障礙 已回復之功能，在，，流控ON，，202區段時，所有前端系統的請求到達入口層 ιοί後即馬上回應前端訊息:”xx功能現無法服務請稍後再試丨，，，不會進入 ©整p層102甚至到後端系統達到流控目的。透過此設計系統可自動進入流 控狀態，並可自動偵測發生障礙之服務介面是否已修復，當一有成功回應 之資料後，該介面流控狀態馬上解除，進入正常服務流程中。另外，還有 一個決定的條件為系統所設定之Timeout時間值〇(單位為秒），與入口層 執行緒最大個數C(MaxThread C_)，錄τ與L錢根财運轉狀況 來設定，如果前端請求量很大，而發生障礙的服務介面遲遲無法修復，那 麼鎖死在平台上之請求量相對也會很多，因此需符合〇L/2T<c，即發生帳 礙進入流控時最壞的狀態下，入口層1〇1之執行序數需大於瞬間之可能測 藝试數。實際入口層1〇1執行序數當然要比此測試數多，才有辦法服務其它 正常之服務。 、 (3)自動分派功能： 在多主機所組合而成之大型企業整合平台上，如何發揮多主機的優勢 協同合作創造最佳效能是本發明目標。該入口層1〇1、整合層1〇2與轉換層 1〇3間均以非同步之訊息仵列(Message Queue)溝通，而轉換層1〇3與基層 ⑽是與後端系統所提供介面規約溝通(如HTTp、s〇Ap、MQ、s〇CKET)， 透過轉換層102再將不同格式與編碼資料轉換為uTF-8之XML格式資料， 並透過訊息佇列與上一層模組溝通。整合平台上透過不同規約之入口層1〇1 提供不同之服務介面功能(即透過單一入口介面即可使用所有後端系統所提 200947322 供之服務）’經由管理者介面為不同服務功能訂定服務名稱， 如”L1.XXX.REQ”,L1指的是入口層服務，XXX代表不同服務名稱，針對每 一個’’L1.XXX.REQ”會相對在訊息伺服軟體上設定一 L1XXXQ訊息佇 列；而每一部硬體伺服器上都有一 u包含於整合層1〇2中) 之軟體模組隨時在讀取此一訊息佇列(U.XXXQ)上之資訊，那一部伺服器 先s賣取資料就執行該服務内容(服務内容是定義在一 SERVICE XML描述檔 中）’在SERVICE.XML中該服務所定義之<F〇RK_LIST一SUBCMD>，如果 有多筆，表示此服務要同時整合轉換層1〇3多個子服務，因此，會並行產 生多筆佇列(L2.YYY.QJL2.ZZZ.Q…)資料到轉換層1〇3，等待轉換層1〇3回 ®應，在轉換層103回應後整合層1〇2同步各子服務内容後透過入口層1〇1 -回應給前端系統。同樣的轉換層1〇3中每個子服務也有各別定義之訊息佇 列L2.YYY_Q，與L2.YYY_Listener之軟體模組，隨時等待讀取此一訊息佇 列(lixxx-Q)上之資訊’那一部伺服器先讀取資料就執行該服務内容(服務 内容是定義在一 SERVICE.XML描述擋中），個別子服務會經由轉換規約與 .資料格式然後透過後端系統介接以取得資料，並再透過仔列回應給整合層 102,整合層1〇2依據Service.xmL中之定義收集所有子服務回應資料後， 並加以整併回應給前端系統(如圖三所示）》由此設計整合平台上之所有硬體 ❹伺服器所執行的内容是一致的，當某一台伺服器上之整合平台負載比較 輕，便會優先抓取入口層101所寫入的資料來執行，當所有硬體效能不足 時可以很容易的加入硬體，以提升效能，係如圖四所示，因為所有伺服器 所執行模組與設定内容是一致的。 本發明之企業資訊整合服務分派與流控系統及其方法與習知技術相 較之下，更包含以下優點： 1·本發明透過系統線上自動偵測模組，並記錄及運算所有線上介面最新 運作狀況，當線上應用介面服務發生問題時，除一般告警外，系統並會自 動進入控管狀態’不至於因單-系統或少數介面出現問題而拖個服務 平台。 11 200947322 2_本發明係提供整合平台線上流量自動分派管理功能，自動分派前端應 用系統加諸於本平台之負載。 3.本發明係將不同系統所提供之應用介面透過此一平台整理與轉換後 能夠將不一致之通訊規約與資料格式，轉換為一致之規約與資料格式，如 此前端應用系統不再因為與後端系統間之連接模式不同而耗費研發人力與 時程》 、 4·本發明係透過四個層次(基層、轉換層、整合層與入口層)之軟體架 構’將各別系統所提供之介面(基層）由不同通訊規約(Pr〇t〇c〇l)與資料内容轉 換，成為中間層並可透過組裝將現有介面組成更大的介面資料(整合層），最 ·〇後’再次轉換為提供服務之入口層，提供單一資料介面格式與Pr〇t〇e〇i給前 端應用系統使用。 5.本發明係包含後端應用系統介面狀態監測功能、後端應用系統流量 控管功能、可有效控管企業應用軟體整合及後端應用系統，使其能在異常 情況發生的同時，即可適時自動將異常的狀況排除不必透過系統管理人 '員，達到企業應用系統整合全年365天無休的運作。 、上列詳細說明乃針對本發明之—可行實施舰行具舰明，惟該實施 例並非用以限制本發明之專利範圍，凡未脫離本發明技藝精神所為之等效 _實施或變更，均應包含於本案之專利範圍中。 综上崎’本案不僅於技術思紅確顧新，並具備習狀傳統方法 力讀合购似妙性之找㈣專利要 申喝恕明貝局核准本件發明專利申請案，以勵發明，至 激I {v»便。 【圖式簡單說明】 圖A本發狀型企業資訊整合服務分派與流控系統及其方法之架構 圖二為本發明大型企業 區間流控之流程圖。 資訊整合服務分派與流控系統及其方法之時間 12 200947322 圖三為本發明大型企業資訊整合服務分派與流控系統及其方法之多層 次服務規約專換之架構圖； 圖四為本發明大型企業資訊整合服務分派與流控系統及其方法之多層 次服務分派與硬體主機關係之架構圖。 【主要元件符號說明】 101入口層 102整合層 103轉換層 104基層 ·© 105管理模組200947322 Pills, invention description: [Technical field of invention] The present invention relates to an enterprise information system integration service distribution and flow control method, in particular for an application in an enterprise information service integration platform architecture, providing dynamic monitoring of each back-end application system Function and service dispatch and flow control method functions. [Prior Art] In order to solve the problem of different system interfaces and data dispersion, large enterprises have developed the requirements of Enterprise Application Integration or Service Oriented Architecture, enterprise application software. The main purpose of the integration is to, ©, integrate the internal applications of the enterprise through the enterprise serviee bus, to achieve information transparency, and the enterprise can use the information immediately and effectively, and internal management and decision-making Various commercial business competition occasions. In order to enhance the market advantage, the marketing and telecom business promotion of new business can achieve instant messaging, information integration and process automation, create greater profitability in existing resources, rely on rapid internal resource integration and rapid Integration and provision of customer data and services. However, the existing enterprise application software integration technology, while providing the function of real-time information integration under normal circumstances, fails to provide the ability to provide error φ prevention or automatic flow control in exceptional cases. When a part of the back-end system connected to the integrated platform has a problem, and the front-end system user (Call Center, Internet Web Site, IVR) cannot wait for a long time, and _ keeps making requests, if the integration platform does not provide one The ability of error detection and flow control will eventually cause problems in the system due to exhaustion of resources. Most of today's system analysis problems only end on the online-recording (l〇g) function. It is impossible to judge the abnormality of the information flow in real time, and make decisions and control in the system immediately without the intervention of system maintenance personnel. For example, enterprise application software integration can provide user identification to filter illegal users, but it does not prevent malicious users from malicious or unintentional damage. If the user does not send out messages continuously, the system will slow down. 'Or resources are occupied and cannot be used by other legitimate and normal users. Another example is that a back-end application system does not work properly, causing the information flow to stay in the enterprise application software integration system 5 200947322, causing the entire integrated platform system to be dragged down, resulting in no large amount of coffee and manpower, and then finding out the _ Managers often spend and let system administrators run out. Therefore, in order to make the enterprise gallery use large protection costs. Work, need to have a set, can be traced to the tree on the various aspects of the face of the secrets of the secret can be efficient Taiwan does not affect the operation of other systems due to the situation of the various systems: the system is more operational, to maximize effectiveness. The application of software m is lost, and the analysis of each pure plum on the real 3 good = ^, the tree is lacking in a good design, and needs to be improved. New, ίίΓΓί Lin on the wealth of the Wei Wei points, g think to improve the creation of service points _ π force research and development, corporate information [invention content] This is the purpose of providing an enterprise-level software integration service platform The automatic control, through the system = line soil automatic _ Huan' and record the operation status of the interface on the Yunlin tree, when the online application media account occurs _, in addition to the general information, green will automatically enter the control state , not to drag a service platform due to problems with a single or a few interfaces. Luben's first goal is to provide an integrated platform online traffic automatic dispatch management function that automatically distributes the load applied to the platform by the terminal application system. The third object of the present invention is to convert the unconformed communication protocol and data format into a consistent protocol and data format by using the platform provided by the same software. The system no longer spends research and development manpower and time course because of the different connection mode with the back-end system. The fourth object of the present invention is to convert the interfaces (base layers) provided by the respective systems from different communication protocols (Protocd) and data content through the software architecture of the four layers (base layer, translation layer, integration layer and entry layer). Become the middle layer and assemble the existing interface into a larger interface data (integration layer). Finally, it is converted into the entry layer of the service, providing a single data interface format and Pr〇t〇c〇l 200947322 to the front end. Application system use. Miscellaneous and lion-like light-hearted integration services distribute money (four) and its methods, providing an integrated service platform for enterprise automation distribution and flow control, including people with different statutes (-) π layer Huan (4) s〇Ap, Hm Toilets, squatting, and various stipulations can be assigned to all back-end systems to form an enterprise-level virtual database. More complex functions may require integration of more than one interface. Provide more 70 services 'System administrators use the browser to set the flow control parameters. The invention also includes an integrated platform system instant flow control module for monitoring and flow control of an application service executing the integrated platform. ' ' The invention also provides a complete and integrated information for application within the enterprise, between various application systems. _ Exchange platform architecture 'to achieve the system - control and service interface, then the focus of the database system, host system ( VAX 'IBM host, HP host) and master-slave architecture can provide integrated services via this integrated platform system. The front-end system dispatches a module by calling the entry layer of the present invention, and the module has the function of collecting the execution state of the interface, and dispatching the front-end request to the integration layer module function, and the integration layer module determines whether to integrate one according to the content of the request. The above sub-services, while simultaneously requesting multiple conversion layers, each conversion layer module receives the request of the integration layer module, and then converts the protocol _〇甽 and the data format to the base request, where the base layer refers directly Connect the original interface module provided by the back end φ, such as the telephone company's local call acceptance green interface, the city scam system interface, the mobile business accounting system interface, and so on. Each request will be recorded, and the status of the query will be stored in the data immediately, and each read will be returned to the import layer and the front-end application system for identification and further flow control processing. The present invention also provides a self-discipline distribution architecture. The main part is that the management front-end system accesses the back-end system through the service integration platform of the present invention. Through the entry layer, the request result and state of each front-end are transmitted through the multi-level distribution module. After being processed and judged by the entry layer, the management module records the processing status. When any service function has a problem, the service enters the flow control through the entry layer. [Embodiment] 200947322 Please refer to the figure - as shown in the figure, the structure diagram of the company's enterprise-owned women's joint venture (four) control system and its method, the structure includes: an entry layer 101 'The entrance layer 101 provides a single protocol and义^Data format interface entry' and build a multi-level dispatch module and instant flow control module at the entry level 101; - Integration layer 1〇2' The integration layer 1〇2 integrates different sub-services into a large service, and An integration module is built in the integration layer 102; a conversion layer 1 〇 3 'the conversion layer 103 converts different protocols and data formats into a unified format' and a conversion module is built in the conversion layer 103; a base layer 104, the base layer The 104-series application and the back-end system are compliant with the back-end system; © a management module 105, which sets system parameters. - Through the above structure, the following functions can be provided: (1) Avoid deadlock (DeadLock) problems caused by synchronous applications; (2) Traffic management functions; (3) Automatic dispatch function. - The methods for achieving the above functions are described below: (1) Avoid deadlock problems caused by synchronous applications: The Enterprise Service Integration Messaging Platform provides different interface technologies through different protocols ( Protocol) The conversion between assets and the 'integration of the system, the integration of different service interfaces, can be woven in the enterprise Weidu money towel _ anal, so that the county (four) code (s_e Code) in the system The business logic can be separated, and the endpoint-to-endpoint processing ((4) in End Pro(10)) function is adopted. The large-scale enterprise considers the information flow of the Laibu, and avoids the closure (Dead 1 magnetic) riding caused by the synchronous application. . The enterprise f service integration provides four single service interface layers (HTTP, SGAP, MS, SQCKET), and the single-media® can access all functions of the system. (2) Traffic management function: Through the service interface status (calculate the application time and response of each application interface on the integrated platform), each user interface can get the response value of ___, which can be the previous day average response 200947322 Time or monthly average value, which can be obtained by calculating the L〇g data in the database. Of course, because each service interface has different operational contents in each back-end system (such as some are simple query databases, Some may be different, different hardware and software performance is different), therefore, the average time value of the running response calculated by each interface is not the same. The so-called traffic management is an expected value f(Mn#) that can be accepted by a certain service interface set by the running response average time (Mn seconds, Μη represents the average response time value of the nth interface), and the expected value is The system can tolerate the longest response time of the service interface. When a service interface response time exceeds f(Mn)B, the entry interface layer records the value (c(n)=c(n)+l c(8) for the nth service interface to continuously exceed the monitored value f(Mn) The number of times, here is the dynamic record using the Hash table method. ❹ The entry layer 101 module reads the flow control threshold value τ(η) set by the manager interface for -/; IL control (when T(n) > c(n)), the service interface is immediately When entering the flow control, all requests from the front-end system for this service will be rejected by the entry layer 1〇1 module to reduce the abnormal load of the integrated platform. After the service interface enters the flow control, it is very important to automatically restore the operation of the service interface after the backend resumes normal operation. The general design is to start another test module to periodically perform the cycle for the interface, but This will be a problem, if it is a different service interface, it will be difficult to handle (because as long as a test will write the data into the real database on the line or into a complex transaction process), such as applying for an ADSL door number, in- In a complex organization, it is impossible to implement a test interface for each φ transaction service interface on the host. Therefore, a time interval flow control module is designed in this integration platform to solve this problem. A large enterprise often has hundreds or even thousands of front-ends. 'Integrated platforms may also have thousands of application service interfaces running at the same time. Each of the bribes can be used as a big bribe. When there is a problem with the function, such as responding to the time report or even not responding (locking), hundreds or even thousands of requests are locked on the integrated platform in an instant. If there is no good mechanism to solve this problem automatically, it is not just the former. The user system waits for the response result for a long time, and the resources of the integrated platform (cpu, Mem〇ry, I/〇) may exhaust the software and hardware resources due to the abnormal operation of a few service functions, so that other normal services; Face-to-face service can't be used for front-end secrets, so in addition to the use of the Thread group control system resources, and timely alerting the target system that has been blocked (through 200947322 via email, newsletter, etc.) ) 'The need to further apply the time interval flow control module for control. Referring to FIG. 2, the present invention represents a service interface function c(8) > τ(8). The function has entered the time interval flow control, and the τ value can be set to any value by the management interface, but usually with the backend system repair and interface time. Relevant, in general, in units of divisions, such as 5 seconds, test (10), catch time • k's absolute shot end request to reach the entry layer 1〇1 immediately respond to the front-end message:,, this service interface is abnormal Unable to service, please try again later. At this time, the number of requests for the corpse and the management interface can be reached to the backend system through the entry layer 101. Therefore, in addition to the flow control, the function of the test backend is blocked. In the flow control ON, 202 segment, all front-end system requests will reach the entry layer ιοί and immediately respond to the front-end message: "xx function is now unavailable, please try again later,, will not enter © full p Layer 102 even reaches the backend system for flow control purposes. Through this design system, the flow control state can be automatically entered, and the service interface of the obstacle is automatically detected. When the data is successfully responded, the flow control state of the interface is immediately released and enters the normal service flow. In addition, there is a decision condition for the Timeout time value (in seconds) set by the system, and the maximum number of threads in the entry layer C (MaxThread C_), which is set by the operation status of τ and L money. The amount of front-end requests is large, and the service interface with obstacles cannot be repaired. Therefore, the number of requests for locking on the platform will be relatively large. Therefore, it is necessary to comply with 〇L/2T<c, which is the most troublesome when entering the flow control. In the bad state, the execution number of the entry layer 1〇1 needs to be larger than the instantaneous possible test number. The actual number of executions of the actual entry layer 1〇1 is of course greater than this number of tests, and there are ways to serve other normal services. (3) Automatic dispatch function: How to exert the advantages of multi-host on a large enterprise integration platform composed of multiple hosts Collaboration and cooperation to create optimal performance is the object of the present invention. The entry layer 1〇1, the integration layer 1〇2 and the conversion layer 1〇3 communicate with each other via a non-synchronized message queue (Message Queue), and the conversion layer 1〇3 and the base layer (10) are interfaces provided with the backend system. Protocol communication (such as HTTp, s〇Ap, MQ, s〇CKET), through the conversion layer 102, converts different formats and encoded data into uTF-8 XML format data, and communicates with the previous module through the message queue. On the integration platform, different service interface functions are provided through the different levels of the portal layer 1.1 (that is, the services provided by all back-end systems can be used through a single portal interface). 'Setting services for different service functions through the administrator interface Name, such as "L1.XXX.REQ", L1 refers to the entry layer service, XXX represents different service names, and for each ''L1.XXX.REQ'), a L1XXXQ message queue is set relative to the message servo software; And each hardware server has a software module included in the integration layer 1〇2) to read the information on the message queue (U.XXXQ) at any time, and the server first s The service content is executed by selling the information (the service content is defined in a SERVICE XML description file) 'The SERVICE.XML defines the <F〇RK_LIST-SUBCMD> defined by the service. If there are multiple pens, the service is required to be At the same time, more than one sub-services of the conversion layer are integrated. Therefore, multiple columns (L2.YYY.QJL2.ZZZ.Q...) data are generated in parallel to the conversion layer 1〇3, waiting for the conversion layer 1〇3 back®, After the conversion layer 103 responds, the integration layer 1〇2 synchronizes the sub-services. After the content is transmitted to the front-end system through the entry layer 1〇1. Each sub-service in the same conversion layer 1〇3 also has a separately defined message queue L2.YYY_Q, and the software module of L2.YYY_Listener is ready to be read at any time. The information on this message queue (lixxx-Q) 'that server reads the data to execute the service content (the service content is defined in a SERVICE.XML description block), and the individual sub-services will pass the conversion protocol. And the data format is then interfaced through the backend system to obtain the data, and then responded to the integration layer 102 through the serialization, the integration layer 1〇2 collects all the sub-service response data according to the definition in Service.xmL, and then integrates Responding to the front-end system (shown in Figure 3), the content of all the hardware and software servers designed on the integrated platform is consistent. When the integrated platform on a server is lightly loaded, it will take precedence. Grab the data written by the entry layer 101 to perform, and when all the hardware performance is insufficient, the hardware can be easily added to improve the performance, as shown in Figure 4, because all the modules and settings of the server are executed. The enterprise information integration service dispatching and flow control system and the method thereof of the present invention have the following advantages compared with the prior art: 1. The invention automatically detects the module through the system line, and records and calculates all the operations. The latest operation of the online interface, when there is a problem with the online application interface service, in addition to the general alarm, the system will automatically enter the control state 'not to drag a service platform due to single-system or a few interfaces. 11 200947322 2_ The invention provides an automatic distribution and management function of online traffic on the integrated platform, and automatically allocates the load of the front-end application system to the platform. 3. The invention converts the application interface provided by different systems through the platform and converts the inconsistent communication protocol and data format into a consistent protocol and data format, so that the front-end application system is no longer because of the back end. The connection mode between systems is different and the R&D manpower and time course are spent. 4. The invention provides the interface provided by each system through the software architecture of the four levels (base layer, translation layer, integration layer and entry layer). ) Converted from different communication protocols (Pr〇t〇c〇l) and data content into intermediate layers and can assemble existing interfaces into larger interface materials (integration layer) through assembly, and then re-convert to provide services The entry layer provides a single data interface format and Pr〇t〇e〇i for use by front-end applications. 5. The invention comprises a back-end application system interface state monitoring function, a back-end application system flow control function, an effective control enterprise application software integration and a back-end application system, so that the abnormal situation can occur at the same time Automatically eliminate abnormal conditions in a timely manner without having to go through the system administrator's staff to achieve enterprise application system integration 365 days a year. The above detailed description is for the purpose of the present invention, and the embodiment is not intended to limit the scope of the invention, and the equivalents or implementations are not deviated from the spirit of the invention. It should be included in the patent scope of this case. In summary, this case is not only in the case of technical thinking, but also in the traditional way of reading the purchase of the genius. (4) The patent must apply for the application of the invention patent application by the Mingbei Bureau to encourage the invention. Inspire I {v». [Simple diagram of the diagram] Figure A The architecture of the information integration service dispatching and flow control system and its method in the present invention Figure 2 is a flow chart of the interval flow control of the large enterprise in the present invention. Time for information integration service distribution and flow control system and its method 12 200947322 Figure 3 is a structural diagram of the multi-level service specification replacement of the large-scale enterprise information integration service distribution and flow control system and its method in the present invention; Enterprise information integration service distribution and flow control system and its method of multi-level service distribution and hardware host relationship architecture diagram. [Main component symbol description] 101 entry layer 102 integration layer 103 conversion layer 104 base layer · © 105 management module
1313