ࡱ> 130bjbjUU >&??|| 7999999.L99N77MO#d0zzz099z| : Resumo: Nas reas da fsica, biologia, ecologia, entre outras, existe a necessidade crescente de executar experimentos que requerem ambientes de processamento de alto desempenho, por tratarem de aplicaes computacionalmente complexas ou por envolverem um grande volume de dados. Esses experimentos, no contexto de e-Science, so formados por inmeras tarefas ou aplicaes que normalmente so executadas diversas vezes e respeitam uma sequncia de execuo pr-determinada. Esses experimentos so frequentemente denominados como workflows cientficos. Devido complexidade dessas aplicaes, geralmente necessrio unir ambientes computacionais de alto desempenho para atender as necessidades dos workflows, por exemplo, para aumentar a capacidade de processamento, formando ambientes heterogneos e compartilhados. Dado que o problema de escalonamento de tarefas em ambientes homogneos considerado da classe NP-Completo, o problema de escalonamento de workflows cientficos em diferentes ambientes computacionais com sistemas de gerenciamento de tarefas distintos, formado por mquinas heterogneas e sem garantia de exclusividade, torna a obteno de um escalonamento eficiente ainda mais difcil por tratar de aplicaes intrinsecamente complexas e de grande porte, em ambientes compartilhados com comportamentos pouco previsveis. Outro ponto a ser considerado o gerenciador de tarefas de cada ambiente, que so independentes e nem sempre oferecem apoio a workflows cientficos. Alm disso, na maioria das vezes, esses gerenciadores so centralizados, uma abordagem que tornase pouco eficiente para o gerenciamento de quantidades significativas de workflows e de recursos em grande escalas. Nesse contexto, diferentemente dos sistemas gerenciadores atuais, esse trabalho prope a criao de uma abordagem de escalonamento e gerenciamento descentralizado de workflows cientficos, portvel para diferentes e-infraestruturas, respeitando as polticas de acesso de cada ambiente e permitindo requisies simultneas de mltiplos usurios. Mais especificamente, a abordagem proposta tem por objetivo adaptar um Sistema de Gerenciamento de Aplicao (SGA) para criar workflows cientficos autnomos que consigam se autogerenciar e explorar as caractersticas intrnsecas de aplicaes e-Science e dos recursos alocados, visando otimizar a utilizao das e-infraestruturas para reduzir o tempo de execuo dos workflows cientficos. Enquanto o SGA no necessita ser instalado nos recursos que sero utilizados, tambm oferece mecanismos de tolerncia a falhas, escalonamento de tarefas e balanceamento de carga dinmico. Desta forma, possvel executar workflows em ambientes que antes no tinham suporte para workflows. Essa abordagem parece ser mais escalvel do que as abordagens existentes e mais capacitada para encarar as principais caractersticas dos ambientes computacionais atuais que so dinmicos (com variao na capacidade e na disponibilidade), instveis e no exclusivos (recursos compartilhados entre diferentes usurios). Palavras-chave: Computao autnoma; Workflows cientficos; Escalonamento dinmico e distribudo; Sistemas de gerenciamento; E-infraestruturas. Abstract: In scientific fields of study like physics, biology, ecology, among others, there is a growing need to perform experiments that require high performance computing environments because the research involves computationally complex applications or large amounts of data. In the context of e-Science, these experiments, often referred to as scientific workflows, are typically made up of numerous tasks that are executed several times while respecting a predetermined execution sequence. With the increasing complexity of these applications, scientists face difficulties obtaining access to sufficient computational resources. It may even be necessary to aggregate distinct computational environments to address the needs of these workflows. More often than not, the most commonly avaliable e-infrastructures are heterogeneous and shared. Given that the classic task scheduling problem, even in homogeneous environments, is considered to be NP-Complete, the problem of efficiently scheduling multiple scientific workflows in different computational environments, formed by heterogeneous and shared machines, is even more difficult given one must deal with intrinsically complex applications running in large scale environments whose job management systems often operate independently and generally do not support scientific workflows. If one considers that nearly all such management systems are centralized, it is not hard to see that this traditional approach will become extremely inefficient when having to manage significant numbers of workflows and resources. In this context, unlike current management systems, this work proposes the adoption of a decentralized scheduling and management approach for scientific workflows one that is portable to different e-infrastructures, respects the access policies of each environment and allows simultaneous requests from multiple users. More specifically, the proposed approach aims to adapt an existing Application Management System (AMS) to create autonomous scientific workflows that can self-manage and exploit the intrinsic characteristics of its e-Science applications and the available resources, in order to optimize the use of the e-infrastructure and to reduce the execution time of the workflow. While the AMS does not need to be pre-installed on the resources, it also provides mechanisms for fault tolerance, task scheduling, and dynamic load balancing. In this way, it is possible to harness e-infrastructures that previously did not support workflows. This approach seems to be more scalable than existing ones and better able to address the main characteristics of todays computational environments that are dynamic (varying in capacity and availability), unstable and shared amongst multiple users. Keywords: Autonomous Computing; Scientific Workflows; Distributed and dynamic Scheduling; Management systems; E-infrastructure.   = F   S\vjsFGQR]^gp㽮hdh\mH sH h\mH sH h ^h\mH sH h\5CJ\aJmH sH "h ^h\5CJ\aJmH sH h ^h\5\hdh\6]h\h ^h\5CJ\aJh\5CJ\aJ8  6 7 ~_`$a$gd ^gd ^&'45:;GP[\_`vjkZcrs _`ih ^h\5\mH sH hdh\mH sH h\mH sH h ^h\mH sH ?<P1h:p\. A!"#$% Dp^ 666666666vvvvvvvvv666666>6666666666666666666666666666666666666666666666666hH6666666666666666666666666666666666666666666666666666666666666666662 0@P`p2( 0@P`p 0@P`p 0@P`p 0@P`p 0@P`p 0@P`p8XV~_HmHnHsHtH@`@ NormalCJ_HaJmHsHtHDA D 0Default Paragraph FontRiR 0 Table Normal4 l4a (k ( 0No List PK![Content_Types].xmlj0Eжr(΢Iw},-j4 wP-t#bΙ{UTU^hd}㨫)*1P' ^W0)T9<l#$yi};~@(Hu* Dנz/0ǰ $ X3aZ,D0j~3߶b~i>3\`?/[G\!-Rk.sԻ..a濭?PK!֧6 _rels/.relsj0 }Q%v/C/}(h"O = C?hv=Ʌ%[xp{۵_Pѣ<1H0ORBdJE4b$q_6LR7`0̞O,En7Lib/SeеPK!kytheme/theme/themeManager.xml M @}w7c(EbˮCAǠҟ7՛K Y, e.|,H,lxɴIsQ}#Ր ֵ+!,^$j=GW)E+& 8PK!Ptheme/theme/theme1.xmlYOo6w toc'vuر-MniP@I}úama[إ4:lЯGRX^6؊>$ !)O^rC$y@/yH*񄴽)޵߻UDb`}"qۋJחX^)I`nEp)liV[]1M<OP6r=zgbIguSebORD۫qu gZo~ٺlAplxpT0+[}`jzAV2Fi@qv֬5\|ʜ̭NleXdsjcs7f W+Ն7`g ȘJj|h(KD- dXiJ؇(x$( :;˹! I_TS 1?E??ZBΪmU/?~xY'y5g&΋/ɋ>GMGeD3Vq%'#q$8K)fw9:ĵ x}rxwr:\TZaG*y8IjbRc|XŻǿI u3KGnD1NIBs RuK>V.EL+M2#'fi ~V vl{u8zH *:(W☕ ~JTe\O*tHGHY}KNP*ݾ˦TѼ9/#A7qZ$*c?qUnwN%Oi4 =3ڗP 1Pm \\9Mؓ2aD];Yt\[x]}Wr|]g- eW )6-rCSj id DЇAΜIqbJ#x꺃 6k#ASh&ʌt(Q%p%m&]caSl=X\P1Mh9MVdDAaVB[݈fJíP|8 քAV^f Hn- "d>znNJ ة>b&2vKyϼD:,AGm\nziÙ.uχYC6OMf3or$5NHT[XF64T,ќM0E)`#5XY`פ;%1U٥m;R>QD DcpU'&LE/pm%]8firS4d 7y\`JnίI R3U~7+׸#m qBiDi*L69mY&iHE=(K&N!V.KeLDĕ{D vEꦚdeNƟe(MN9ߜR6&3(a/DUz<{ˊYȳV)9Z[4^n5!J?Q3eBoCM m<.vpIYfZY_p[=al-Y}Nc͙ŋ4vfavl'SA8|*u{-ߟ0%M07%<ҍPK! ѐ'theme/theme/_rels/themeManager.xml.relsM 0wooӺ&݈Э5 6?$Q ,.aic21h:qm@RN;d`o7gK(M&$R(.1r'JЊT8V"AȻHu}|$b{P8g/]QAsم(#L[PK-![Content_Types].xmlPK-!֧6 +_rels/.relsPK-!kytheme/theme/themeManager.xmlPK-!Ptheme/theme/theme1.xmlPK-! ѐ' theme/theme/_rels/themeManager.xml.relsPK] & ^rL\d>@@@UnknownG*Ax Times New Roman5Symbol3" ArialACambria Math":SOS%!0$P ^!xxResumo:HelioHelioOh+'0x  4 @ LX`hpResumo:HelioNormal_WordconvHelio1Microsoft Office Outlook@`4<@L@@O%՜.+,0 hp|   Resumo: Title !"#$%&')*+,-./2Root Entry FMO41TablezWordDocument>&SummaryInformation( DocumentSummaryInformation8(CompObjy  F'Microsoft Office Word 97-2003 Document MSWordDocWord.Document.89q  F#Documento do Microsoft Office Word MSWordDocWord.Document.89q