Abstract
End-user oriented system for massive computations in grid is proposed and implemented. The system provides support of user interfaces for input and output data staging, asynchronous jobs submission and control, tasks status and results monitoring. The main system components include web-portal, authentication components and jobs submitter that interact with grid infrastructure. The main advantages of described grid-portal are flexibility in computations back-ends support and possibility to interactively handle thousands of jobs. The proposed integrated environment was implemented in Ukrainian National Grid (U NG) infrastructure for massive simulations of non-linear dynamics in neuroscience.
Highlights
Grid [1] is one of the most popular tools for large scale high-throughput computing
Compared to the old clients where grid job ID was the ultimate source of information, a new clients require additional info to operate with jobs
ARC 1.x comes with built-in support for accessing Storage Resource Manager (SRM) compatiple storage elements and LCG File Catalog (LFC) data catalogue services, which are de-facto standards in the global grid infrastructures like EGI and WLCG
Summary
Grid [1] is one of the most popular tools for large scale high-throughput computing. Ukrainian gridinfrastructure was established in 2006 as Ukrainian Academic Grid (UAG) [2]. The manual operation requires a lot of knowledge beyond the subject of researcher’s science area, like internals of grid operation, job description languages, command line client tools usage for job submission, monitoring, management and data staging. Human factor is another disadvantage of the manual operation. Special parallel software for neurons dynamics modelling was developed by authors of this paper [4] This software was deployed for grid-computing with integrated environment for handling massive job submission. In present works this approach is used for building of grid-portal for nonlinear dynamics simulations in neuroscience
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have