Speaker
Lev Shamardin
(Scobeltsyn Institute of Nuclear Physics, Moscow State University (SINP MSU))
Description
Grid systems are used for calculations and data processing in various applied
areas such as biomedicine, nanotechnology and materials science, cosmophysics
and high energy physics as well as in a number of industrial and commercial
areas. Traditional method of execution of jobs in grid is running jobs directly
on the cluster nodes. This limits the choice of the operational environment to
the operating system of the node and also does not allow to enforce resource
sharing policies or jobs isolation nor guarantee minimal level of available
system resources. We propose a new approach to running jobs on the cluster nodes
when each grid job runs in its own virtual environment. This allows to use
different operating systems for different jobs on the same nodes in cluster,
provides better isolation between running jobs and allows to enforce resource
sharing policies. The implementation of the proposed approach was made in the
framework of gLite middleware of the EGEE/WLCG project and was successfully
tested in SINP MSU. The implementation is transparent for the grid user and
allows to submit binaries compiled for other operating systems using exactly the
same interface from the standard gLite user interface node. Virtual machine
images with the standard gLite worker node software and sample MS Windows
execution environment were created.
| Presentation type (oral | poster) | oral |
|---|
Author
Lev Shamardin
(Scobeltsyn Institute of Nuclear Physics, Moscow State University (SINP MSU))
Co-authors
Alexander Kryukov
(Scobeltsyn Institute of Nuclear Physics, Moscow State University (SINP MSU))
Andrey Demichev
(Scobeltsyn Institute of Nuclear Physics, Moscow State University (SINP MSU))
Ilya Gorbunov
(Scobeltsyn Institute of Nuclear Physics, Moscow State University (SINP MSU))
Slava Ilyin
(Scobeltsyn Institute of Nuclear Physics, Moscow State University (SINP MSU))
