ABLE 2.0.0 07/02/2003 10:25:01

Package com.ibm.able.autotune

AutoTune is an agent-based approach to automated tuning that does not require prior knowledge of the controlled system that is being tuned.

See:
          Description

Interface Summary
AutotuneAdaptor The AutotuneAdaptor interface defines an AbleBean that plays the role of an adaptor for an AutotuneAgent to an application or target system.
AutotuneAgentIf  
AutotuneController The AutotuneController interface defines an AbleBean that plays the role of a controller in an Autotune Agent.
AutotuneDataBean  
 

Class Summary
Autotune  
AutotuneAgent Base Autotune Agent for Generic Adaptive Control
AutotuneAgentBeanInfo  
AutotuneAgentCustomizer This class provides the customizer interface for the AutotuneAgentCustomizer.
AutotuneAgentDataBean  
AutotuneAgentMetricPanel This class provides a panel for the Autotune Agent Customizer.
AutotuneAgentPanel This class provides a panel for the Autotune Agent.
AutotuneAgentTargetPanel This class provides a panel for the Autotune Agent Customizer.
AutotuneConfigurationMetric  
AutotuneMetric An AutotuneMetric defines a single parameter on a target system It can represent a configuration settings, a workload indicator, a service level , or a tuning control Metrics can have sensors/effectors associated with them to do the actual method calls on the target system Or, their values can be set by the process() method on the Adaptor or Controller
AutotuneMetricCollection This class contains a collection of AutotuneMetrics The metrics can be retrieved in a single Hashtable, or as a set of 4 Vectors based on metric type The entire collection of metrics can also be logged to an external file It is meant to be used as a convenience class by AutotuneAgents and Controllers
AutotuneMetricManager The Autotune Metric Manager provides a unified view of all target system metrics to the controllers in the Autotune agent.
AutotuneMetricOperator This class performs binary relation tests against 2 AutotuneMetricValues The result is either a boolean or a double (delta) value use the test() method to check a boolean relation use the computeDelta() method to compute a difference
AutotuneMetricTarget  
AutotuneRemotePanel This class displays a modal dialog for the selection of a remote agent.
AutotuneServiceLevelMetric  
AutotuneTuningControlMetric  
AutotuneWorkloadMetric  
BaseAutotuneAdaptor  
BaseAutotuneAdaptorCustomizer This class provides the customizer interface for the BaseAutotuneAdaptor.
BaseAutotuneAdaptorMetricPanel This class provides a panel for the Autotune Agent Customizer.
BaseAutotuneAdaptorTuningPanel This class provides a panel for the Autotune Agent Customizer.
BaseAutotuneController  
BaseAutotuneControllerCustomizer This class provides the customizer for the Basic Neural Autotune Controller.
BaseAutotuneControllerMetricPanel This class provides a panel for the Autotune Agent Customizer.
BasicNeuralAutotuneController This class implements a basic neural controller It features a neural system model that is trained with off-line data and a neural controller that is adapted on-line The model can predict one or more service level metrics given a set of configuration, workload, service level metrics and tuning controls The controller can specify one or more tuning control values given a set of configuration, workload, service levels, and current tuning control metrics How to use this controller ....
BasicNeuralAutotuneControllerBeanInfo  
BasicNeuralAutotuneControllerCustomizer This class provides the customizer for the Basic Neural Autotune Controller.
BasicNeuralAutotuneControllerPanel This class provides the panel for the Neural Controller tab.
Fuzzy2WayLoadBalanceController This autotune controller uses fuzzy rules to balance the load between two service level metrics, using a single control value
Fuzzy2WayLoadBalanceControllerBeanInfo  
Fuzzy2WayLoadBalanceControllerCustomizer This class provides the customizer for the Basic Fuzzy Autotune Controller.
Fuzzy2WayLoadBalanceControllerPanel This class provides a panel for the Autotune Fuzzy Controller.
Neural2WayLoadBalanceController This class implements a basic neural controller It features a neural system model that is trained with off-line data and a neural controller that is adapted on-line The model can predict one or more service level metrics given a set of configuration, workload, service level metrics and tuning controls The controller can specify one or more tuning control values given a set of configuration, workload, service levels, and current tuning control metrics How to use this controller ....
Neural2WayLoadBalanceControllerBeanInfo  
Neural2WayLoadBalanceControllerCustomizer This class provides the customizer for the Neural 2Way Load Balance Controller.
Neural2WayLoadBalanceControllerPanel This class provides a panel for the Autotune Fuzzy Controller.
TestAutotuneAdaptor  
TestAutotuneAdaptorBeanInfo  
TestAutotuneAdaptorCustomizer This class provides the customizer interface for the BaseAutotuneAdaptor.
 

Package com.ibm.able.autotune Description

AutoTune is an agent-based approach to automated tuning that does not require prior knowledge of the controlled system that is being tuned.

An AutoTune enabled target system exposes metrics for workloads (e.g., RPC arrival rates), configuration (e.g., processor speeds), and service levels (e.g., response times) as well as a means to manipulate tuning controls (e.g., admission control parameters). Our approach is to construct a generic model of the target system (e.g., by training a neural network) and from this derive a controller.


ABLE 2.0.0 07/02/2003 10:25:01

(C) Copyright IBM Corporation 1999, 2003