Speaker
Mr
Omair Shafiq
(DERI Innsbruck)
Description
Over the last few years, Grid Computing has adapted to Web
Services
standards with the emergence of Web Services Resource
Framework (WSRF).
This alignment with Web Services standards has made the Grid
standards more
scalable by enabling Grid resources interoperable over the
Web. The next step
for Grid is Semantic Grid in which knowledge about resources
is exposed and
handled explicitly.
It is envisioned as semantic layering over the current Grid
infrastructure.
Semantic Web Services technologies can be applied to the
Grid in a way to
achieve the Semantic Grid vision. For example, the
background knowledge and
vocabulary of a Grid middleware component could be captured
in ontologies.
Metadata can be used to label Grid resources and entities
with concepts, for
example describing a data file in terms of the application
domain where it is
used. Rules and classification-based reasoning mechanisms
could be used to
generate new metadata from existing metadata, for example
describing the
rules for membership of a VO and reasoning that a potential
member’s
credentials are satisfactory.
Moreover, activities like Grid service discovery or
negotiation of service level
agreements, among others, can be potentially enhanced using the
functionalities provided by Semantic Web Service technologies.
Started far apart in applications and technology, grid
services and web services
converged (Grid services started with GT1, web services
started with XML,
SOAP and WSDL, and finally converged as WSRF, an extension
of web services
that considers grid specific requirements), but still they
are defined at a
syntactical level, without any formal
semantic that would make the suitable for automation. A
technology is needed
that will extend and enrich these service descriptions with
new elements that
will allow to automatize service related tasks. The
conceptual model of WSMO,
and WSML, the formal language that reflects the conceptual
model, will allow
different types of automatic proofs to be made in this
language, thus providing
a certain level of automation for service related tasks.
The OGSA framework, the conceptual model for grids, defines
different types of
services with specific capabilities that are needed for grid
applications.
However, OGSA doesn’t provide a formal language for
describing these
services, thus being of little use in automatic performance
of different service
related tasks. Current languages used in
grid, like GRAM (grid resource allocation manager), are
based on XML and XML-
schema, thus inheriting all its drawbacks (semi structured
data format, no
formal semantics, no reasoning support, etc); here is where
OGSA could benefit
from the conceptual model of WSMO and its associated
language, WSML.
All the OGSA services, which are summarized below could
employ WSMO for
semantically describing their properties: Infrastructure
Services - OGSA
leverages Web services architecture to implement a SOA. WSDL
is used for
service descriptions. SOAP is the communication protocol.
Here WSMO/WSMX
will be helpful. Initially this infrastructure was based on
OGSI but now it will be
based on WSRF.
WSMO conceptual model can be extended to cope up with
complexity of Grid
Services. The extended WSMO model for Grid can act as
specification to realize
Semantic Grid
Services.
Execution Management Services - These services deal with the
problems of
task initiation and management. Grid resource broker uses
the status
information for each resource stored in monitoring and
discovery service to
discover a resource and initiate a job or set of jobs on
that resource. After
scheduling jobs, it also gathers information on the status
of jobs. With the
emergence of Semantic Grid, the information stored by Grid
resource broker has
to be made semantically annotated which will enable the Semantic
Grid infrastructure to manage, monitor and discover the
available grid
resources automatically, more easily and more precisely.
Data Services - Data services in the grid are responsible
for efficient data
access, data consistency, data persistency, data integration
and data location
management.
Triple Space Computing associated closely with Semantic Web
Services can
bring next level of advancement in the Data services of gird.
Resource Management Services - These services allow the
management of
individual resource itself, management of resources in Grid
(i.e. resource
reservation, monitoring
and control) and monitoring of Grid infrastructure which
consists of resources
as well like monitoring the registry service. Semantic
annotations to grid
services will improve the Resource Management by providing
explicit meta-
information to describe the resources of Grid.
Security Services - Security services provide controlled
access to resources
which can be in various administrative domains with
different access and
security policies. Grid Security Infrastructure (GSI)
specifies the whole set of
protocols and security architectures that are required for
controlled resource
sharing across the Grid. Semantic annotations to Grid
resources and other
entities will also help the security services to have a
clear and unambiguous
scenario and to act accordingly for authentication and
authorization of the
resources.
Self-Management Services - Self-management services include
SLA, policies and
service level manager model. SLA includes business and IT
agreements
between the provider and user of the service. Policies are
used to govern the
behavior of an SLM (Service Level Manager) and the
manageable resources
under its control. Service Level Manager Model provides the
interface such that
various human operators and SLM can work together without
having
knowledge about each other built in at design time. Service
level agreements
(SLAs) are currently being investigated by Semantic Web
Services community as
well. However, explicit availability of meta information
will enable the self
management of services more automatic and dynamic.
Information Services - Monitoring and Discovery Service in
OGSA is an XML
database called Xindice that collects information from each
of the resource in
Grid. It Stores dynamic data for monitoring and discovery.
It also acts a registry
for grid services. XPath and XQuery languages are used to
query data from
information service. Semantic annotations to the
intermediate information of
Grid will make the monitoring and discovery in Grid more
precise and accurate in
complex and dynamic scenarios.
Acknowledgements
The work is funded by the FIT-IT (Forschung, Innovation,
Technologie -
Informationstechnologie) under project GRISINO - Grid
semantics and intelligent
objects. The authors would like to thank all the people who
are involved in
GRISINO project and the funding support from Austrian
Government.
Summary
In this extended abstract (poster) we have provided a brief
overview that how
Semantic Web Service technologies can be applied to Grid
standards to realize
the vision of Semantic Grid.
This is an on-going work under Austrian funded project GRISINO (Grid
Semantics and Intelligent Objects) where we plan to realize a
conceptual model
for Semantic Grid Services based on our experiences and expertise
of WSMO as
Semantic Web Services conceptual model. The work will further be
extended to
investigate new requirements for WSML to formally describe Grid
Services and
Jobs.
Author
Mr
Omair Shafiq
(DERI Innsbruck)
Co-author
Mr
Ioan Toma
(DERI Innsbruck)
