An industry blueprint showing how software, hardware and
standards work together in autonomic computing systems is expected
to be discussed during IBM's developerWorks Live conference next
week in New Orleans.
The company also detailed four technologies it has developed
related to autonomic computing, its name for computing systems that
adjust to workloads, identify, predict and solve problems and
require less human intervention.
The blueprint will show customers and developers how to assemble
products from a range of suppliers to create a "self-managing"
autonomic system.
"The industry really needs an overall approach to the
broad-based goals we've described in autonomic computing," said
Alan Ganek, IBM vice-president for autonomic computing.
"You have to have the blueprint for how all of this comes
together. No one supplier can provide all of the pieces, so our
customers need to see how all of the pieces fit together."
The blueprint represents a "major advance" for developers, he
said. The blueprint describes what IBM refers to as "the control
loop", which analyses, monitors and makes changes in autonomic
systems, connecting all the possible components within such a
system, such as software applications, servers, storage, databases
and middleware.
The four specific technologies IBM also announced are:
- Log and Trace tool for problem determination, which helps to
take autonomic systems from figuring out the problems to debugging
applications and middleware.
- ABLE (Agent Building and Learning Environment) Rules Engine for
Complex Analysis, which uses a set of algorithms that allows
intelligent agents to capture data, and can predict future steps to
take based on system experience.
- Monitoring Engine, which is available in IBM Tivoli Monitoring
now, enables root cause analysis for IT failures, server-level
correlation of multiple IT systems and automated corrective
measures.
- Business Workload Management for Heterogeneous Environments,
which will be out in IBM Tivoli Monitoring for Transaction
Performance version 5.2 in the second half of this year, uses the
ARM (Application Resource Measurement) standard to determine why
bottlenecks happen, using response time measurement, transaction
processing segment reporting and learning of transaction workflow
through servers and middleware. The software then adjusts resources
to avoid bottlenecks.
Figuring out how and why problems occur is a big vexation for IT
staff, so IBM is focusing this round of autonomic announcements on
what it calls "problem determination". With existing computer
systems, IT staff are left to identify problems and try to find a
fix for them after, say, a sudden and unexpected server workload
surge.
Using autonomic computing, IT staff would establish performance
objectives for the systems they oversee.
In the case of an e-business infrastructure behind a website,
"one of the key things you want to monitor is what is the response
time and what is the web traffic coming in", said Ric Telford, IBM
director of architecture for autonomic computing.
"If the response time slips below a performance level based on a
large volume of transactions coming in, that's something you need
to take action on."
In the autonomic world, the computer system will do that without
human intervention before the spike occurs and response time slows.
It will adjust automatically to add more servers to handle
workload.
The blueprint also includes the emerging OGSA (Open Grid Systems
Architecture) standard. IBM is one of the companies working on that
standard.