IBM has released early versions of three grid-based
technologies intended to help corporate and third-party developers
better manage grid-based resources in order to solve large and
compute-intensive problems.
The products collectively give developers access to idle
computing capacity, improve their productivity, and more smoothly
transform business processes as a way to extract more value from IT
investments.
"The idea behind making these still-emerging technologies
available now is to get the market involved. It gives us a chance
to get feedback from the market, which helps us shape these
technologies. And it gives us a chance to gain acceptance in some
key areas," said Amit Patel, emerging-technology strategist for
alphaWorks.
The first, called ZetaGrid, helps developers who are strapped
for compute power to access the free capacity of dozens of
computers, both local and remote.
IBM said it is the first specification of a J2EE interface that
helps to "split up" an application and send it to multiple sites
within a grid environment. With the tool, programmers can work
simultaneously on complex development projects or problem-solve
within an existing one.
The technology is also meant to offer a more efficient
infrastructure for protecting grid resources. IBM has already
tested the technology in its Boeblingen labs to verify a
mathematical computation across a grid of thousands of computers in
a heterogeneous environment.
"Right now it is being used by almost 11,000 computers in 70
different countries that are trying to solve the Reyman Hypothesis,
a very large and well-known math problem attempting to verify the
zeroes of the zeta function," Patel said.
A second technology, Grid File Replication Manager, which
improves on the existing open source-based Replica Location Service
from Globus, is a web-based tool for managing grid replicas.
In working with dozens of development partners who are widely
dispersed geographically, it is often necessary to create remote
and read-only copies, or replicas, of files. Replication can be
used to reduce the latency commonly associated with accessing
files.
"This can make it easier to manage replicas of data on the grid.
It also gives you an easier way to create and search through
entries in the (Globus) Replica Location Service," said Patel.
The third technology is Distributed Parallel Programming
Environment for Java, a set of tools for compute-intensive
applications, mainly in the fields of bio-informatics as well as
financial analytics, which often require the creation of markedly
faster distributed, and parallel applications.
"This [technology] introduced the concept of distributed threads
so developers can concentrate their programming on applications
that require these threads. It provides the infrastructure support
for the distribution and execution of those threads as well as
message passing and synchronisation. Essentially, it addresses the
issue of performance in a highly computational environment," Patel
explained.
The new technologies will support Open Grid Services
Architecture, which is a set of specifications and standards that
serves as a common framework for building grid applications.
Ed Scannell writes for Infoworld