RLCatalyst Auto-Remediation Engine
As enterprises deal with lot of connected devices and tools , there is a need to automate the task of monitoring and tracking these resources . This paves way for creating intelligent or self aware resources which can do self-checks, self-correction and self-control . RLCatalyst provides a platform that supports and drives self-awareness.
RLCatalyst offers a Self aware system which can create self aware elements like nodes, jobs, services etc to sense changes, deviations / anomalies from normal behaviour through a sensory mechanism and take remediation steps to get back to normal state. A set of rules define the system response on adverse events. These rules are used by the systems to control the factors that cause adverse events, thus enabling efficiency, availability and responsiveness.
Here is the solution overview:
1. Deploy a monitoring solution that checks various drivers for system performance like CPU, memory, disk, network, applications. Measure health as well operational metrics
2. A time series database to persist health and operational metrics of the monitored systems
3. Rules/configurations to define the behaviour and actions to be taken in the event of abnormal behaviour
4. A notification mechanism to notify users in real time
Remediation Engine to act in the event of adverse / abnormal behaviour based on the rules defined
Everything in nature has some form of self awareness
Self Aware elements in computing can be basic building blocks or can be made of multiple self aware elements
Four drivers for self aware systems
– Operate efficiently with no service disruptions
• Self Preservation
– Act to avoid loosing resources
• Resource Acquisition
– Act to acquire resources
– Find new ways to increase utility
Anything that can be measured can be controlled
When an event occurs, its condition can be checked against known values / states and if there is an anomaly or deviation an action can be taken to correct the condition. In a sense it is a feedback loop. RLCatalyst Pulse is designed to observe events, check if result of events pass a condition, if not create an action to meet the desired condition.
Various methodologies exist achieve this and we propose to use MAPE-K methodology invented and suggested by IBM.
Using the methodology MAPE-K ( Monitor, Analyser, Planner, Executor and domain knowledge) we have an autonomous system, that gets launched, monitor its environment, analyse the monitored data, plan its course correction and execute the correction plan.