Academic Journal
Инструментарий децентрализованного обслуживания потоков параллельных MPI-задач в пространственно-распределенных мультикластерных вычислительных системах
| Τίτλος: | Инструментарий децентрализованного обслуживания потоков параллельных MPI-задач в пространственно-распределенных мультикластерных вычислительных системах |
|---|---|
| Πηγή: | Вестник Томского государственного университета. Управление, вычислительная техника и информатика. |
| Στοιχεία εκδότη: | Федеральное государственное бюджетное образовательное учреждение высшего профессионального образования «Национальный исследовательский Томский государственный университет», 2011. |
| Έτος έκδοσης: | 2011 |
| Θεματικοί όροι: | ДИСПЕТЧЕРИЗАЦИЯ ПАРАЛЛЕЛЬНЫХ ПРОГРАММ, ПРОСТРАНСТВЕННО-РАСПРЕДЕЛЕННЫЕ ВЫЧИСЛИТЕЛЬНЫЕ СИСТЕМЫ, GRID-СИСТЕМЫ |
| Περιγραφή: | Parallel program scheduling refers to the meaningful problem of computer systems (CS) functioning organization. For every incoming program the execution computing resources (subsystems) have to be found. System composition dynamics and variable workload of computer nodes should be taken into account by job scheduler. Centralized schedulers have the inherent significant shortcoming: CSs front-end failure may cause a whole system failure. Besides that, time overhead for resources search increase in case of using these tools in large-scale CS. Thus problem of development of decentralized models, algorithms and system software for parallel program scheduling in distributed CS is urgent. In case of decentralized scheduling collective of schedulers have to function in system and search necessary resources for programs. It allows reaching persistence also in large-scale CS, making them able to continue functioning at some subsystems failures. Decentralized algorithms and software of parallel program scheduling in geographicallydistributed and GRID systems are proposed in this paper. Geographically-distributed CS, composed of H subsystems, is considering. N is total number of computer nodes on subsystems. Let ni be number of nodes in subsystem i S = {1, 2, …, H}. Let also bij = b(i, j, m) be channel capacity between i, j S subsystems while transferring m bytes message ([b(i, j, m)] = byte/s). Scheduler functioning on each subsystem manages parallel program queue and searches for computing resources for job execution. Collective of schedulers is represented by graph G = (S, E), where schedulers and logical links between them correspond to vertices and edges. Presence of (i, j) E edge means that scheduler i is able to send resource requests (jobs of its queue) to scheduler j. Set of vertices j incident to i represents its local neighborhood L(i) = {j S: (i, j) E}. User sends job with resource request to one of the schedulers. That scheduler (in accordance with realized algorithms) searches (sub)optimal subsystem j∗ S for users job to compute. Let job is characterized by rank r number of parallel branches, expected execution time t (walltime) and total size z of execution files and data ([z] = byte). GBroker software suite of parallel MPI-programs decentralized scheduling in geographicallydistributed CS has been created and is being developed in the Centre of Parallel Computing Technology of Siberian State University of Telecommunications and Information Sciences (CPCT SibSUTIS) and Computer Systems Laboratory of Institute of Semiconductor Physics of SB RAS (ISP SB RAS). Software suite includes scheduler gbroker, client tool gclient and netmon monitor of channels performance between subsystems at TCP/IP protocol level. Developed toolkit of parallel job scheduling was investigated on active multicluster CS, created by CPCT SibSUTIS together with ISP SB RAS. Experiments results have shown that mean job service times in centralized and decentralized scheduling are comparable. Scheduling time is rather minor in comparison with job execution time. Decentralized scheduling software suite is one of necessary components for providing persistence of geographically-distributed multicluster CS of CPCT SibSUTIS and ISP SB RAS. Рассматриваются децентрализованные алгоритмы и программы обслуживания потоков параллельных задач в пространственно-распределенных вычислительных системах (ВС). Описывается функциональная структура разработанного программного пакета GBroker децентрализованной диспетчеризации параллельных MPI-программ в пространственно-распределенных мультикластерных ВС. Приводятся результаты моделирования алгоритмов и оценка их эффективности при реализации на мультикластерных ВС. |
| Τύπος εγγράφου: | Article |
| Περιγραφή αρχείου: | text/html |
| Γλώσσα: | Russian |
| ISSN: | 2311-2085 1998-8605 |
| Σύνδεσμος πρόσβασης: | http://cyberleninka.ru/article/n/instrumentariy-detsentralizovannogo-obsluzhivaniya-potokov-parallelnyh-mpi-zadach-v-prostranstvenno-raspredelennyh-multiklasternyh http://cyberleninka.ru/article_covers/14012464.png |
| Αριθμός Καταχώρησης: | edsair.od......2806..5edd7b8b448d8f86a2384ca7319c6fee |
| Βάση Δεδομένων: | OpenAIRE |
| ISSN: | 23112085 19988605 |
|---|