Goal #7: Grid Infrastructure
Mission:
Work with GriPhyN, PPDG, EU DataGrid, iVDGL and other Grid
projects to ensure that the US and global network infrastructures are
satisfying the programmatic needs of scientists using grids. This goal focuses
on providing dynamic monitoring information to the Grid middleware components
and the Grid applications.
Contact
Contacts: Iosif Legrand snd Dantong Yu
Contact Email:
Iosif.Legrand@cern.ch and
dtyu@bnl.gov
Activities:
- End-to-end performance monitoring.
We use existing
monitoring modules and design new ones to provide the information necessary to
describe large scale, interconnected computing facilities as well as the
network traffic and performance. The goal is to provide the monitoring
information from these distributed systems to any other decision services in
the grid middleware in a flexible self-describing way. Many Grid application
and middleware needs the network monitoring services, for example, Resource
brokers needs the network bandwidth to schedule the job and data to the
optimized location efficiently. Grid Data Catalog needs the network weather
information to deploy data into multiple locations in advance to reduce the
network traffic on backbone. GridFtp needs the monitoring information to
dynamically adjust the data transfer behavior. The system is based on a service
oriented architecture of loosely coupled components capable to discover each
other dynamically and to interact using self describing data
structures.
- High-performance Grid data transport protocol and data
replication service. Integrate different network-related Grid middle-wares to
satisfy Grid User requirement on data services.
- Bridge between Grid infrastructure and network
research/engineer/infrastructure. We will come up a list of network
requirements on which the Grid infrastructure is based to provide satisfying
services for Grid users. We will also work with network engineers to understand
how the network could be efficiently used, and provide the list of the most
important things that the network engineers wish grid users/developers should
know. Then we will work with Grid communities to make sure that these concerns
should be taken into consideration during Grid infrastructure design and
deployment.
- Integration of system and network monitoring data with
application monitoring tools.
Milestones:
- Deploy MDS monitoring for certain experiments. (Done)
- Provide web interface for accessing MDS monitor system.
(Done: url:
http://www.atlasgrid.bnl.gov/testbed/SC2002/gridmonitoring.php).
- Deploy MonaLisa with snmp support and Ganglia interface to
CMS sites ( Done at Fermilab, Caltech, UCSD, UFL and CERN ).
- Deploy Ganglia monitoring tools in several US computing
facilities. (In progress url: http://www.atlasgrid.bnl.gov/ganglia/
).
- Work on documents on network requirements which Grid
infrastructure needs to provide success grid services. (Winter 2002).
- Work on Network Bridge plan to include top concerns that the
network engineers wish grid experts should know. (Winter 2002).
- Develop a way to gather data from network weather services
and publish into grid monitoring infrastructure. (Winter 2002).
- Integration prototypes of monitoring services with decion
services in grid middleware and application.
Presentations:
- A prototype for a Monitoring Service slides (ppt) I. Legrand
, May 2002, Fermilab 2.
- A Distributed Agent-based Architecture for Dynamic Services
paper (pdf) slides (pdf) Harvey B. Newman, Iosif C. Legrand, Julian J. Bunn
CHEP - 2001, Beijing, Sept 2001.
- "Atlas
Monitoring Infrastructure in Grid Environment" presented by Rich Baker at
the PPDG collaboration meeting Toronto Feb 2002.
- "ATLAS
Grid" presented by Rich Baker at the PPDG collaboration meeting Toronto Feb
2002.
Links and Related Activities:
|
