第一篇：云計(jì)算不是一種服務(wù)是一項(xiàng)技術(shù)

云計(jì)算不是一種服務(wù)是一項(xiàng)技術(shù)

在2011CENC中國(guó)企業(yè)網(wǎng)絡(luò)通信大會(huì)上，工信部電信研究院總工程師蔣林濤表示云不是一種服務(wù)或者服務(wù)理念，而是一種技術(shù)，而且不是一項(xiàng)全新的技術(shù)，是一系列技術(shù)的繼承和發(fā)展。

云計(jì)算與三種服務(wù)形態(tài)沒(méi)有直接關(guān)系

提到云計(jì)算，人們總是要列出IaaS、PaaS、SaaS這三種服務(wù)形態(tài)。但是對(duì)此，蔣林濤認(rèn)為這三種個(gè)服務(wù)形態(tài)早已經(jīng)存在，不是只有云才能提供，這三種服務(wù)形態(tài)和云計(jì)算并沒(méi)有直接關(guān)系。

“一臺(tái)大型計(jì)算機(jī)外接遠(yuǎn)程終端也可以提供IaaS、PaaS、SaaS這三種服務(wù)。所以從嚴(yán)格意義上說(shuō)，云與服務(wù)沒(méi)有關(guān)系，云是一種實(shí)現(xiàn)技術(shù)，它將分離的IT資源有序的組織起來(lái)和有效的使用起來(lái)。云的用途不會(huì)局限于提供應(yīng)用，在很多平臺(tái)的建設(shè)中也可以采用云計(jì)算技術(shù)，當(dāng)然只有在建設(shè)規(guī)模以上的平臺(tái)體系時(shí)才值得去做，否則僅僅是徒有虛名，而不會(huì)得到實(shí)質(zhì)意義的好處?！笔Y林濤警惕道。

云計(jì)算是IDC的一種補(bǔ)充

目前，云的最主要用途是用于建設(shè)新一代的IDC。IDC是存在已久的IT基礎(chǔ)設(shè)施，IDC一般有兩方面的能力：設(shè)備托管和IT資源出租。其中，IT資源出租就和云計(jì)算有著緊密的聯(lián)系。據(jù)了解，以資源出租為主的IDC有三種工作模式：集中方式、分散方式和分布方式。

蔣林濤認(rèn)為分布式IDC在進(jìn)入云計(jì)算前是沒(méi)有的，分布式是目前IDC缺乏的一種工作模式，而云計(jì)算正好是一種分布式技術(shù)，所以，使用云計(jì)算技術(shù)來(lái)實(shí)現(xiàn)IDC，正好補(bǔ)充了分布式IDC這種工作形式。

（飛象網(wǎng)）

第二篇：服務(wù)是一種幸福

服務(wù)是一種幸福

她，在窗口服務(wù)崗位上，總能讓客戶滿意而歸，用心架起了企業(yè)與客戶的橋梁；她，通過(guò)不斷刻苦努力，掌握了扎實(shí)的業(yè)務(wù)技能，多次被榮獲各級(jí)“服務(wù)明星”榮譽(yù)稱號(hào)；她，只是平凡崗位上的一名普通的收樣員，8年的青春獻(xiàn)給了企業(yè)卻無(wú)怨無(wú)悔。她就是，一位樂(lè)觀上進(jìn)、富有朝氣的80后女孩。2005年，她以天使般的微笑、細(xì)致的服務(wù)和熱情的工作態(tài)度，不僅在平凡崗位上充分體現(xiàn)了自身價(jià)值和獨(dú)特青春風(fēng)采，更完美詮釋了“客戶滿意，我的責(zé)任”的 “陽(yáng)光”服務(wù)理念，贏得了客戶與同事的一致好評(píng)。她說(shuō)，客戶的肯定，是最大的滿足；客戶的理解，是工作的強(qiáng)大動(dòng)力；客戶的抱怨是金，她希望在抱怨中變得更加完美?？蛻艨渌白钯N心”“大伯，千萬(wàn)百計(jì)為客戶省錢(qián)、省心，做好客戶用電的貼心小顧問(wèn)，許多客戶都對(duì)王憬的服務(wù)工作印象深刻。“雖然這的每個(gè)營(yíng)業(yè)員服務(wù)態(tài)度都很好，但我每次到營(yíng)業(yè)廳辦業(yè)務(wù)還是喜歡找小王，因?yàn)樗k事特別用心負(fù)責(zé)，服務(wù)也非常周到，我都把當(dāng)自家的親女兒了！”一說(shuō)起王憬，家住營(yíng)業(yè)廳附近的鄭大娘對(duì)她的服務(wù)工作更是贊不絕口。鄭大娘說(shuō)，她每個(gè)月都會(huì)去營(yíng)業(yè)廳查電費(fèi)，但由于自己沒(méi)讀過(guò)什么書(shū)，不識(shí)字，所以每次去營(yíng)業(yè)廳辦業(yè)務(wù)都很麻煩，也很不好意思，不過(guò)王憬每次都對(duì)她很耐心，不僅會(huì)把電費(fèi)單上的信息一字一句地讀給她聽(tīng)，而且在得知鄭大娘腿腳不便后，更是每月直接將電費(fèi)清單送到她家里，這令鄭大娘感動(dòng)不已，并直夸王憬比自己的兒女還要親。在同事們的眼中，王憬是一個(gè)聰明而且學(xué)習(xí)能力極強(qiáng)的人。她的同事應(yīng)婷說(shuō)，學(xué)習(xí)同樣一個(gè)東西，她總是要比大家要學(xué)的快，同樣的工作任務(wù)，她也總是第一個(gè)出色完成，經(jīng)過(guò)短短幾年的學(xué)習(xí)積累，王憬已經(jīng)掌握了扎實(shí)的業(yè)務(wù)知識(shí)技能，成為了營(yíng)業(yè)廳里的崗位能手，并很快被提拔為營(yíng)業(yè)廳號(hào)長(zhǎng)。因?yàn)橥蹉娇偸窃敢夂屯聜兎窒碜约旱囊恍W(xué)習(xí)、工作經(jīng)驗(yàn)，幫助大家共同提高成長(zhǎng)。新進(jìn)員工吳婧說(shuō)，自新員工進(jìn)入營(yíng)業(yè)廳以來(lái)，是王憬手把手地傳授她們業(yè)務(wù)知識(shí)和技能。而且無(wú)論王憬有多忙，只要有問(wèn)題找她，她都會(huì)非常耐心地進(jìn)行解答，幾個(gè)新人的業(yè)務(wù)水平因此得到了飛速提升，每個(gè)人都已經(jīng)能夠獨(dú)挑大梁。除了佩服王憬工作能力強(qiáng)外，同事們其實(shí)更看中她對(duì)大伙在生活中的那份珍貴情誼。禮儀督導(dǎo)師李衛(wèi)平回憶說(shuō)，有一年王憬過(guò)生日，全家人圍著桌子等她回去慶生，但是家人等了很久卻不見(jiàn)她的到來(lái)，打她的電話也不接，著急的家人只好打電話到單位詢問(wèn)，才知道因?yàn)橥峦蝗簧o(wú)人照看，王憬得知消息后急忙陪她去醫(yī)院打點(diǎn)滴，把手機(jī)落在了辦公室里，直到晚上10點(diǎn)多回單位才看到手機(jī)上幾十個(gè)未接來(lái)電，還有全體同事給她發(fā)的短信：生日快樂(lè)！領(lǐng)導(dǎo)愛(ài)她“敢創(chuàng)新”做好服務(wù)工作需要?jiǎng)?chuàng)新，而王憬就非常喜歡在創(chuàng)新方面動(dòng)點(diǎn)腦經(jīng)，因此部門(mén)領(lǐng)導(dǎo)都很喜愛(ài)她，直夸她是營(yíng)業(yè)廳的參謀長(zhǎng)。更令人沒(méi)想到的是，王憬還根據(jù)不同情境研究出了一套能迅速與客戶搭建信任關(guān)系的溝通技巧：遇到抽煙的客戶，她會(huì)滿臉微笑地提醒他：“您好，請(qǐng)吃顆糖，為了你的健康，請(qǐng)不要吸煙”；代客戶復(fù)印身份證，細(xì)心地她發(fā)現(xiàn)當(dāng)天是客戶生日，會(huì)送上誠(chéng)懇地祝福：“祝您生日快樂(lè)”；在為客戶辦理免費(fèi)短信提醒，發(fā)現(xiàn)對(duì)方手機(jī)號(hào)碼很有規(guī)律，她會(huì)說(shuō)：“您的號(hào)碼很好，真好記”。此外，她會(huì)給每位來(lái)辦理業(yè)務(wù)的客戶發(fā)放名片，并建立了一個(gè)電力客戶QQ群，以便為客戶提供多渠道的服務(wù)。家人想她“早回家”“我們一個(gè)大辦公室30多號(hào)人，她常常是最晚一個(gè)回家?！焙芏嗳苏J(rèn)為電力窗口服務(wù)工作枯燥乏味，她卻有不同見(jiàn)解：“把簡(jiǎn)單的事情重復(fù)千萬(wàn)次地做好，就是不簡(jiǎn)單；把容易的事情認(rèn)真地做對(duì)，就是不容易?！闭f(shuō)到底就是把一件事情重復(fù)千百次的做對(duì)了，但是這對(duì)于很多人來(lái)說(shuō)，就是一個(gè)活生生的奇跡，而她卻做到了。這就是王憬，明星班組中一顆閃耀的明星，她說(shuō)，服務(wù)工作是美麗的，在“陽(yáng)光”營(yíng)業(yè)廳工作就是她此生最大的幸福。

第三篇：跆拳道是一項(xiàng)運(yùn)用手足技術(shù)

岳輝跆拳道暑假班招生簡(jiǎn)章

跆拳道是一項(xiàng)以腿為主、以拳為輔的競(jìng)技運(yùn)動(dòng)，被稱為“踢的藝術(shù)”。主要學(xué)習(xí)內(nèi)容分為腿法、品勢(shì)（拳法）、禮儀三大部分。通過(guò)日常學(xué)習(xí)，增進(jìn)腿法，品勢(shì)的技藝，提高學(xué)員耐力、柔韌、爆發(fā)和協(xié)調(diào)等身體素質(zhì)。同時(shí)，輔以“禮儀廉恥、忍耐克己、百折不屈”的學(xué)習(xí)精神，使學(xué)員在健體同時(shí)強(qiáng)志，懂得忍耐和尊重。

醫(yī)學(xué)博士麥克羅依先生指出練跆拳道有如下益處：可刺激和開(kāi)發(fā)練習(xí)者的速度、力量，靈敏耐力等身體素質(zhì)?？商岣咝难芟到y(tǒng)、呼吸循環(huán)系統(tǒng)機(jī)能以及中樞系統(tǒng)的靈活性，可克服自身的懶惰、消極、怯懦、畏懼等不良因素?？膳囵B(yǎng)堅(jiān)韌果斷，自信良好的意志品質(zhì)，從而造就健康的身心等。健體：健體作用不言而喻。其動(dòng)作中的各種踢腿、手臂的上格、下截、拉伸等動(dòng)作，可使肌肉和力量得到增強(qiáng)，肌腱、韌帶、肌肉的彈性得以提高，從而增強(qiáng)人的體力和柔韌性。啟智： 跆拳道可給大腦許多良好刺激。練習(xí)過(guò)程不單是骨骼肌肉的收縮，而是空間感知、經(jīng)驗(yàn)、類型識(shí)別等右腦功能的綜合發(fā)揮，因此非常有利于發(fā)展青少年的空間轉(zhuǎn)換、形體感知等右腦功能，提高了形象思維和創(chuàng)造力。育心：跆拳道需要通過(guò)做變換方向的動(dòng)作來(lái)明確方位，從而發(fā)展了空間知覺(jué)；此外，孩子們還要主動(dòng)想辦法克服困難，這鍛煉了意志，培養(yǎng)了勇敢精神，學(xué)會(huì)了對(duì)人、事、物的正確態(tài)度。知禮：跆拳道推崇“禮始禮終”的尚武精神，講究未曾學(xué)藝先學(xué)禮，未曾習(xí)武先習(xí)德。通過(guò)師生間的互敬、活動(dòng)時(shí)的敬禮儀式、同學(xué)之間的相互禮讓等，讓孩子理解了“禮”的含義，知道怎樣做才是“禮”的表現(xiàn)。

一、招生對(duì)象：幼兒園中大班、小學(xué)生、高中生、上班族。

二、教學(xué)內(nèi)容：

1、跆拳道的歷史、特點(diǎn)、精神、禮儀。

2、太極八章前三章。

3、腿法、拳法、步法、身法、防守法。

三、教練——畢業(yè)于西安體育學(xué)院本科學(xué)士，國(guó)家一級(jí)裁判，國(guó)家一級(jí)運(yùn)動(dòng)員，跆拳道教練培訓(xùn)講師，多年跆拳道教學(xué)經(jīng)驗(yàn)。曾多次獲多個(gè)省市冠軍。

四、開(kāi)課時(shí)間學(xué)生放假后的第二天到8月31號(hào)，每天上課時(shí)間90分鐘。

五、課程設(shè)置上午 6：00----7:30

聯(lián)系電話：******

第四篇：云計(jì)算及其安全技術(shù)學(xué)習(xí)心得

云計(jì)算及其安全技術(shù)學(xué)習(xí)心得

作為一個(gè)云計(jì)算技術(shù)研究的工作人員，通過(guò)本課程的學(xué)習(xí)加深了我對(duì)云計(jì)算及其安全技術(shù)的認(rèn)識(shí)。主要的心得體會(huì)如下：

隨著科學(xué)技術(shù)的迅速發(fā)展，現(xiàn)代信息化建設(shè)對(duì)信息量的要求越來(lái)越高，傳統(tǒng)IT技術(shù)已不能滿足要求，高效率、低成本地對(duì)海量數(shù)據(jù)進(jìn)行存儲(chǔ)和處理的云計(jì)算技術(shù)就應(yīng)運(yùn)而生。云計(jì)算[1]是基于互聯(lián)網(wǎng)的相關(guān)服務(wù)的增加、使用和交付模式，通過(guò)互聯(lián)網(wǎng)來(lái)提供動(dòng)態(tài)易擴(kuò)展虛擬化的資源。云計(jì)算將賦予互聯(lián)網(wǎng)更大的內(nèi)涵，更多應(yīng)用能以互聯(lián)網(wǎng)服務(wù)的方式交付和運(yùn)行；同時(shí)也擴(kuò)大了IT軟硬件產(chǎn)品應(yīng)用的外延和改變了軟硬件產(chǎn)品的應(yīng)用模式。云計(jì)算是一種新的商業(yè)模式，帶動(dòng)IT產(chǎn)業(yè)格局的變化，為全球信息技術(shù)產(chǎn)業(yè)帶來(lái)全新的發(fā)展機(jī)遇，同時(shí)也深刻影響世界經(jīng)濟(jì)的發(fā)展。

我國(guó)政府高度重視對(duì)云計(jì)算的發(fā)展，把云計(jì)算列為重點(diǎn)發(fā)展的戰(zhàn)略性新興產(chǎn)業(yè)。2012年9月，為了加快推進(jìn)云計(jì)算技術(shù)創(chuàng)新和產(chǎn)業(yè)發(fā)展，科技部發(fā)布了《中國(guó)云科技發(fā)展“十二五”專項(xiàng)規(guī)劃》。權(quán)威機(jī)構(gòu)預(yù)測(cè)，未來(lái)三年中國(guó)云計(jì)算產(chǎn)業(yè)鏈的產(chǎn)值規(guī)模將達(dá)到2000億元?！?013-2017年中國(guó)云計(jì)算產(chǎn)業(yè)發(fā)展前景與投資戰(zhàn)略規(guī)劃分析報(bào)告》指出，近年來(lái)企業(yè)在不斷增加其云服務(wù)資產(chǎn)，并已經(jīng)準(zhǔn)備好開(kāi)始集成工作；調(diào)查發(fā)現(xiàn)超過(guò)74%的企業(yè)客戶計(jì)劃用服務(wù)提供商來(lái)實(shí)現(xiàn)系統(tǒng)集成。同時(shí)，隨著我國(guó)智慧城市建設(shè)的鋪開(kāi)，以及各地方的公有云和大型企業(yè)私有云建設(shè)的快速開(kāi)展，中國(guó)云計(jì)算市場(chǎng)規(guī)模將進(jìn)一步增大。

云計(jì)算技術(shù)的快速發(fā)展，用戶對(duì)其可靠性和安全性有了越來(lái)越多的需求。根據(jù)美國(guó)研究公司Gartner發(fā)布的《云計(jì)算安全風(fēng)險(xiǎn)評(píng)估》的報(bào)告稱云計(jì)算服務(wù)存在著如下七大潛在的安全風(fēng)險(xiǎn)[8]，安全問(wèn)題依然不容忽視：可審查性、特權(quán)用戶的接入、數(shù)據(jù)隔離和加密、數(shù)據(jù)恢復(fù)、數(shù)據(jù)位置、長(zhǎng)期生存性以及調(diào)查支持持特定的調(diào)查。

云計(jì)算安全技術(shù)主要表現(xiàn)在身份的保護(hù)、基礎(chǔ)設(shè)施的保護(hù)和信息數(shù)據(jù)的保護(hù)三個(gè)方面。第一個(gè)方面就是身份的保護(hù)。對(duì)于身份安全來(lái)說(shuō)，用戶需要強(qiáng)認(rèn)證機(jī)制，一般考慮ID和密碼的保護(hù)。在云環(huán)境中需要身份認(rèn)證技術(shù)，才能實(shí)現(xiàn)云服務(wù)和應(yīng)用的安全遷移；第二個(gè)方面是基礎(chǔ)架構(gòu)的安全保護(hù)?；A(chǔ)架構(gòu)主要包括軟硬件、網(wǎng)絡(luò)設(shè)備、操作系統(tǒng)和應(yīng)用環(huán)境等。對(duì)基礎(chǔ)架構(gòu)安全來(lái)說(shuō)，要確?；A(chǔ)架構(gòu)的安全是一項(xiàng)非常大的挑戰(zhàn)。這就需要非常強(qiáng)的可信鏈條，這個(gè)可信鏈條不僅包括硬件安全，還有因特網(wǎng)、操作系統(tǒng)、虛擬化等，整個(gè)鏈條都要有非常強(qiáng)的可信性；第三個(gè)方面就是數(shù)據(jù)和信息的安全保護(hù)。對(duì)于數(shù)據(jù)和信息的安全保護(hù)也有兩個(gè)方面，既數(shù)據(jù)的完整性和信息的保密性。數(shù)據(jù)的完整性主要指數(shù)據(jù)在被訪問(wèn)的時(shí)候，不應(yīng)該被隨意篡改、增加或刪減。信息的保密性指信息只能是有權(quán)限的人才能閱讀。這方面的技術(shù)主要有分布式的密鑰管理技術(shù)、密鑰的加密技術(shù)以及DLP技術(shù)等。

云計(jì)算作為一項(xiàng)新興的技術(shù)，特點(diǎn)突出、優(yōu)勢(shì)明顯，得到了眾多人的追捧。云計(jì)算技術(shù)高速發(fā)展給互聯(lián)網(wǎng)帶來(lái)了新的機(jī)遇與挑戰(zhàn)，各大IT公司加入到云計(jì)算行業(yè)，推出自己的云服務(wù)。但與此同時(shí)，云計(jì)算還面臨諸多安全隱患，尤其是云安全問(wèn)題，有待云計(jì)算相關(guān)安全技術(shù)的進(jìn)一步研究解決。

第五篇：云計(jì)算 服務(wù)遷移service_migration總結(jié)

1.introduction 云服務(wù)遷移是云計(jì)算實(shí)現(xiàn)中的一個(gè)概念，它確保了個(gè)體或組織的數(shù)據(jù)，或其它商業(yè)元素，能輕松地在不同的云供應(yīng)商之間，或在一個(gè)私有云設(shè)施內(nèi)部，轉(zhuǎn)移或遷移，且不用克服實(shí)現(xiàn)、整合、兼容性及互操作性問(wèn)題。通過(guò)服務(wù)遷移，應(yīng)用，基礎(chǔ)設(shè)施，或者任意云主持的應(yīng)用、服務(wù)都能避免被鎖定在一個(gè)運(yùn)營(yíng)商的情況。服務(wù)遷移也定義了應(yīng)用能部署在其它云供應(yīng)商或被私有云架構(gòu)支持的過(guò)程及框架。服務(wù)遷移的整個(gè)過(guò)程由幾個(gè)不同進(jìn)程和技術(shù)組成，它們依賴于被遷移的應(yīng)用或服務(wù)的復(fù)雜性。服務(wù)遷移也包括了通過(guò)在開(kāi)放式標(biāo)準(zhǔn)和框架上設(shè)計(jì)、開(kāi)發(fā)云基礎(chǔ)架構(gòu)，而進(jìn)行的最佳實(shí)踐的開(kāi)發(fā)。

服務(wù)前移是非常復(fù)雜的，因?yàn)榇蠖鄶?shù)云服務(wù)供應(yīng)商使用了專利技術(shù)，為了把數(shù)據(jù)從一個(gè)供應(yīng)商的存儲(chǔ)中轉(zhuǎn)移到另一個(gè)，有時(shí)必須首先把數(shù)據(jù)移回用戶自己的服務(wù)器上，然后再?gòu)倪@里轉(zhuǎn)移到新的供應(yīng)商地點(diǎn)。甚至這樣時(shí)，由于需要格式的變換來(lái)確認(rèn)數(shù)據(jù)也可能需要被重定格式，Even then, the data may need to be reformatted because of formatting changes made to conform to the first service provider’s system.Such a situation may lead to vendor lock-in – a situation in which a product or service vendor has made it so difficult or expensive to transition to a competitor’s offering that customers are effectively “l(fā)ocked in.”

2.Services Migration in Clouds 2.1架構(gòu)

The servicemigration concept extends the notion of a service-orientedarchitecture.Web service has been the major standard for realizingSOA.Web services have good encapsulationand strong integration capabilities.Web services dynamicdistribution.Service providers and service requesters may bedistributed in different geographic locations.There are twoways to complete the request issued by the service requester.One is the data migration.The requester transfers data to aservice provider to have it analyzed.After that, the providerreturns the result back to the requester.The other way is theweb service migration.The requester desires to acquire aservice comes to his system to analyze the data.After thedata is processed, the service is moved from the requester'ssystem.For a large amount of data, such as a large set ofgigabytes to terabytes, rendered across the wide area networkcan be costly, it may increase the communication latencysignificantly.Due to the reason above, many researches work on theweb service migration are interested in minimizing thecommunication cost.In cloud environment,hardwareresources and their management utilities are all provided asservices.A large number of services deployed in differentnetwork domains.So it can directly select the “nearest”servers to migrate the services.The term “nearest” coversessential factors of service quality, including networkbandwidth, network delay, server capacity, service type,connecting fee, etc..A new infrastructure to support themigration of web services in cloud is presented.The proposed infrastructure includes five components: Load&Proximity Analyzer(LPA), Services DependencyAnalyzer(SDA), Publish/Select Manager(PSM), ServiceMigration Manager(SMM), and Service Migration DecisionManager(SMDM).In current web infrastructure, theinteractions among all components are soap messages basedon httpihttps.A host can be a web service provider, a webservice requester, or even a web service proxy.1)負(fù)載&鄰近分析儀

收集所有主機(jī)的cpu負(fù)載信息和主機(jī)與其鄰居間的鄰近信息，并更新維護(hù)。The LPA runs periodically to collect all hosts' cpu loadinformation and the proximity information between the hostand its neighbors.When the network topology changes, itwill result in proximity change.We use the round trip timepackets and the number of hops they had traveled to reflectthe proximity's change.We also may use overage queuingdelay to reflect the cpu's load information.2)服務(wù)依賴分析儀

管理服務(wù)器之間的呼叫依賴性，獲取相關(guān)服務(wù)的配置參數(shù)和它們的依賴性，分析來(lái)確保所選服務(wù)的先前服務(wù)被加入到正確的PSM中。

The SDA is used to manage the calling dependencyamong web services.It's function including capturing theconfiguration parameters of interrelated services and theirdependencies, analyzing the integrity to make sure that theantecedent services of the selected service are added to thecorrect PSM, avoiding incorrect cyclic dependencies ofsystem services, and generating proper configuration settingsequence.3)發(fā)布/選擇管理

如果一個(gè)主機(jī)提供了特定的服務(wù)或接收了從其他主機(jī)遷移來(lái)的服務(wù)，它會(huì)向PSM發(fā)送注冊(cè)請(qǐng)求。

If a host provides a specific web service or receives aweb service which migrated from another host, it will sendthe registration request to the PSM.So there may have manyentries for the same web service.When a requester sendsweb service looking request to the PSM, it should decidewhich entry is the best for the requestor based on therequestor's location, the information from the LPA, and thesecurity factors.4)服務(wù)遷移決策管理

收集分析從LPA何SDA來(lái)的必要數(shù)據(jù)，作出服務(wù)遷移的決定。考慮網(wǎng)絡(luò)帶寬，服務(wù)有效期，服務(wù)器容量，服務(wù)訪問(wèn)頻率，可轉(zhuǎn)移性等。

The SMDM is responsible for collecting and analyzingthe essential data from LPA and SDA to make the decisionon service migration.Factors should be taken into accountare network bandwidth, service valid period, server capacity,service access frequency, and transferability, etc.5)服務(wù)遷移管理

在SMDM決定何時(shí)并向何處遷移服務(wù)后，SMM收集并打包服務(wù)的狀態(tài)信息，把它們轉(zhuǎn)移到目的主機(jī)，并在目的主機(jī)上重建執(zhí)行服務(wù)的環(huán)境。

After the SMDM decides when and where to migrate theweb service, the SMMcollects and packs the states ofservices, transfers them to the destination host, andreconstructs the executing environment on the target host.2.2服務(wù)遷移決策管理機(jī)制

Before SMDM make migration decision, it should takethree aspects into account: whether a web service ismigratable, when to migrate a web service, and where tomigrate a web service.1）In cloud environment, web services not only meansoftware sources, but also include the hardware sources, thedevelopment platforms, the databases, etc.These sources arenot all suitable for migration.For example, during a host isexecuting a web service, it need to access data from adatabase which residents on a remote server.At this time, themigration cost to migrate the database to the host may be toohigh.Even the migration cost is acceptable, the conditions toset up the database may not meet for this host.So we canconsider that the migration of a database is unacceptable.Similarly, it is not feasible to migrate a hardware indistributed area.It denotes these kinds of sources areunmigratable.We can label a web service is migratable orunmigratable when it published.The SMDM may not sendmigration command to the SMM if a web service isunmigratable.2）提出負(fù)載門(mén)檻

When to migrate a web service is an important factor forSMDM to make migration decision.One of the innovationsof this paper is that we propose a schema to determine themigration time.We predefine a overload threshold inSMDM.When the LPA collected the performance data fromall the hosts periodically, SMDM compares the data with theoverload threshold and makes migration decision accordingto the comparison results.For example, we may predefinethe migration condition is cpu's use-ratio more than 80%.Once the data collected from LPA shows a provider's usagerate of cpu is over the threshold, it means the serviceresidents on this provider should be migrated.3）After decided when to migrate, the next step for SMDMis to determine where the appropriate destination server toaccommodate the web service is migrated from the provider.When a provider publish a web service, it also providesconstraints that should be met to execute the web service(such as the required operation system, the minimal memoryrequirement etc)and the priority information among theseconditions.So there may have some candidate hosts in cloudcan provide the appropriate environment to accommodate theweb services.We should choose one among them.Thedecision is based on the conditions and the priorityinformation that the candidate can provide.4）Besides the consideration of constraints and the priorityinformation, the migration cost is also should be taken intoconsideration.When a client issues a web service request toa provider, the SDA constructs the web service call tree, theLPA collect round trip time between hosts and the averagequeuing delay.All these information are sent to the SMDM.SMDM computes the migration cost based on a cost modeland find the best host which yield lowest cost as a target host.3.Agent-based Service Migration Framework in Hybrid Cloud Figure 1 shows the architecture of the proposedsystem, which involves a federated broker and the hybridenvironment.Since the public cloud is charged in pay-perusescheme, in general, it should provide dynamic provisionof resources involving the computing platform and requiredstorage.In other words, the proposed system can start newcomputing cluster from public cloud for new jobsdynamically while the loading of private cloud is overrun.Such capability is provided by most virtualization tools ofvirtual machine, such as XEN, Microsoft’s Hyper-V andKVM.The tools provide Application Programming Interfaceto manage the resource of virtual machine.In addition,private cloud and public cloud can be connected via virtualprivate network(VPN)or public network if do not considerthe security issue between private cloud and public cloud.3.1聯(lián)盟代理

Federated Broker layer is the second tier in thearchitecture.It is responsible for monitoring system resourceand all behaviors of resource, reconfiguring system scale,and migrating services and balancing the load betweenclouds.The layer consists of three agents and tworepositories;there are System Monitoring Agent,Reconfiguration Decision Agent, Service Migration Agent,Profiling Database, and System/Service InformationDatabase, respectively._ System Monitoring Agent(SyMA): The agent isresponsible for collecting the loading andremaindering resource related information of allnodes in the hybrid cloud.It will activelycommunicate with the Cluster Admin Agent(CAA)resided in each cluster periodically to get theinformation or passively receive the heartbeatmessage that contains the information from CAA.The information involving in the heartbeat are asfollows: 1)cluster's name, 2)cluster's loadinformation, 3)jobs etc.Upon receiving a heartbeatmessage, the SyMAwill log the information toProfiling Database and System/Server InformationDatabase._ Reconfiguration Decision Agent(RDA): The RDAis used to reconfigure and adjust the cloudenvironment after the reconfiguration policy isfitted.The RDA can retrieve current cluster's statusand loading information in the cluster to decide toadjust the scale of public cloud or not.If the SyMAis noticed the private cloud is overloading, he willcontacts the RDA to initiate the reconfigurationprocess to balance the loading between private andpublic cloud.Also, the SyMAwill notify the RDAscale the public cloud down if the private cloud isunder loading during a time period._ Service Migration Agent(SeMA): It is responsiblefor managing and initiating the service migrationbetween private and public cloud.If RDA informsit that the public cloud is scaled up and send amessage to migrate a job to public cloud, the SeMAwill initiate the service migration process accordingto the policy of service migration.The policy andalgorithm will be depicted next section._ Profiling Database: The database is to store theprofile of public cloud including the resourceinformation the public cloud has, such as thenumber of VM, the size of memory and storage,the bandwidth of link to public cloud, and theusage of these resources, which can be used to helpfor making decision of scaling._ System/Service Repository: The repository keepsthe system and service related information, such asthe location of private and public cloud, theinformation of services run in the hybrid cloud, andso on.3.2云內(nèi)的代理

Except for the components in the federated layermentioned above, there are two agents will be developedinto both private and public cloud platform.They areCluster Admin Agent and Job Agent._ Cluster Admin Agent(CAA): The CAA is the maincomponent on the cluster, which can manage thewhole cluster and negotiate the actions with agents infederated layer.He also periodically collects cluster’sinformation, and sends them to SyMAthat isembedded within the heartbeat message.Besides, italso manages and guide the running of Job Agents runin the cluster._ Job Agent(JA): In order to deliver a service to othercloud, a service will be encapsulated into a JA beforemigrating to other cloud.A JA has all the informationa service needed, includes service's execution file,service's input data, the location to be migrated, andso on.Code mobility can be divided into two categories [5]:strong mobility and weak mobility.The strong mobility ischaracterized as if it allows executing units to move theircode and execution state to a different site, while weakmobility only allows moving their code.In the currentversion of proposed solution, in order to simplify thecomplexity of current system design, we encapsulate aservice into a JA, and provide weak migration scheme.Thejob model and encapsulating a job into JA are shown inFigure 2.It only restarts the execution instance from thebeginning of the code.In other words, all migrated serviceswill be re-executed in the other cloud environment.3.3服務(wù)遷移流程

圖3是服務(wù)遷移的具體流程，其步驟如下：

1)簇管理代理周期性的發(fā)送包含負(fù)載信息的heartbeat message給系統(tǒng)監(jiān)控代理。2)系統(tǒng)監(jiān)控代理通過(guò)計(jì)算負(fù)載信息來(lái)檢查云的負(fù)載是否平衡。

3)如果非常不平衡，系統(tǒng)監(jiān)控代理通知重配置決策代理調(diào)整公共云的大小。重配置決策代理基于它的決策向公共云的簇管理代理發(fā)送一個(gè)請(qǐng)求。

4)公共云的簇管理代理將啟動(dòng)新的簇來(lái)接受和部署遷移服務(wù)，并把結(jié)果返回給重配置決策代理。5)重配置決策代理接著通知服務(wù)遷移代理遷移服務(wù)至公共云。

6)服務(wù)遷移代理根據(jù)遷移策略計(jì)算哪個(gè)服務(wù)將被遷移至公共云，并通知私有云和公共云。

3.4遷移策略

Although the hybrid cloud is constructed as a federatedmodel, we assume that all services initially are stored in theprivate cloud, excepting exterior job submitted by users.Therefore, all jobs invoked by users are queued in theprivate cloud first.When the load of the private cloudexceeds, the SyMAwill initiate service migration process.服務(wù)遷移策略是在服務(wù)遷移過(guò)程將要開(kāi)始的時(shí)候用來(lái)做出決策。用來(lái)做出決策的算法有3個(gè)，分別是JC(Job Count)，SJ(Size of Job)和EFT(EstimatedFinish Time)。

1)JC方案是最簡(jiǎn)單的一個(gè)。它僅僅根據(jù)兩個(gè)云中作業(yè)的數(shù)量來(lái)做出決定。2)SJ方案基于兩個(gè)云中作業(yè)的總大小來(lái)做出遷移決策。3)EFT方案通過(guò)估算兩個(gè)云中完成作業(yè)的時(shí)間來(lái)做出決策。