Grid(Lab) Resource Management
The aim of this webpage is to describe GridLab Resource Management System (GRMS),
consisting of a set of technical specifications, along with clarifications and examples that help end users, administrators
and developers to deal with this software. This webpage will also provide up-to-date technical guidelines to setup, run and use
the GRMS releases.
GRMS place in the GridLab architecture:
GridLab Resource Management System (GRMS) is an open source meta-scheduling system,
developed under the GridLab Project, which allows developers to build
and deploy resource management systems for large scale distributed computing infrastructures.
GRMS provides developers of user-level functionalities with a more abstract
view of low level and complex Grid technologies. Based on dynamic resource selection and discovery,
mapping and advanced scheduling methodology, combined with a feedback control architecture
and support from other GridLab Middleware Services, it deals with dynamic Grid environment
and resource management challenges, e.g. load-balancing among clusters and various work-load
systems, remote job control or file staging support. Therefore, the main goal of GRMS
is to manage the whole process of remote job submission and control to various
queuing systems (e.g. Condor, PBS, LSF, N1 Grid Engine), clusters systems or resources directly.
GRMS has been designed as an independent set of components for resource management
processes which can take advantage of various low-level Core Services, e.g. GRAM,
GridFTP and Mercury Grid Monitoring System,
as well as various grid middleware services, e.g. GridLab Authorization Service,
Replica Management Service, File Movement Service or Delphoi Services.
All these services working together provide a consistent, adaptive and robust grid middleware layer which fits
dynamically to many different distributing computing infrastructures.
The GRMS as a central middleware service along with other middleware services
improve distributed resource, data and security management (see GRMS 2.0.0 architecture and Middleware Service Layer).
First releases of GRMS are based on a relatively stable Globus 2.4 Toolkit
and uses Globus Core Services deployed on resources, in particular GRAM, GridFTP
and GRIS/GIIS services. Therefore, you need at least GRAM, GridFTP
and GRIS/GIIS services installed and configured properly on all resources
and front-end machines to start playing with GRMS and its basic functionalities
(see GRMS 2.0.0 architecture and Core Service Layer).
From the user point of view the most important thing is the way he can access GRMS.
Due to existing GSI-enabled Web Service interfaces he can get the remote access
using a simple command-line client,
GRMS 2.0.0 internal architecture:
Technically speaking, GRMS is a persistent service within a Tomcat/Axis
environment and it supports Grid Security Infrastructure by providing the GSI-enabled
Web Service interfaces for all clients, e.g. portals or applications,
and thus can be integrated with any other middleware grid environment
(see download section for service and client source code).
Basic components of GRMS have been developed entirely in Java,
and could be installed on various operating systems and resources.
By using GSI-enabled Web Service interfaces GRMS is able to communicate
with portals (an example GRMS portal has been developed based on the open-source
GridSphere portal framework and is available from v1.8.0).
Thus, the access for both end users and administrators can be provided via
graphic web interfaces or even from mobile devices.
A final stable release of GRMS v2.0.0 software together with technical guides and docs are due in November 2004.
Due to the fact that GRMS components are independent from underlying low level solutions
we have started also some tests with Globus 3.0/3.2 mechanisms, Unicore and Web Service interfaces to queuing systems..
However, since WSRF has been announced and widely accepted,
we are keen on moving with GRMS as a brokering service towards
WSRF specifications/solutions in the future .
One of the main assumptions for GRMS is to perform remote
job control and management in the way that satisfies Users (Job Owners)
and Administrators (Resource Owners) and also meets their requirements and preferences.
Therefore, we have introduced within GRMS multi-objective procedures
and optimization techniques to define and build various flexible
resource management strategies.
Taking into consideration all accounting issues we store all actions performed by GRMS
on resources in a database for a further administrative support (e.g. for building fair-sharing
or policy based resource management strategies).
GridLab: Grid Application Toolkit and Testbed
is co-funded by the European Commission under the Fifth Framework Programme
Web admin: Petr Holub, web design: Radoslaw Strugalski
Last update on Tuesday, 05-Apr-2005 15:09:04 CEST.