AIOps & Achieving Full-Stack Service Observability

Observability vs Monitoring

The “Three Pillars” of Observability

Observability enables you to understand what is happening in your IT infrastructure, based on these three data types – the widely accepted as the foundations of observability:

1. Logs: A timestamped, text-based record of an “event”. Logs can be used to create a second-by-second record of an event. Logs are typically what IT teams will examine when a system goes wrong
2. Metrics: Metrics are measures of application and system health over X period of time. Things like CPU usage, run queue duration, memory and so on
3. Traces: A trace is the end-to-end ‘journey’ of every user request, from the UI or mobile app as it courses through the entire architecture and back the way to the user

Enterprises who have adopted tools such as NPM, APM, and network monitoring tooling - generating, in some cases, millions of metrics, logs and traces.

Having access to logs, metrics, and traces doesn’t necessarily make systems more observable.

The Foundations of Observability 

From this position you will gain the ability to leverage your best of breed monitoring tooling, augmenting it in a data-driven fashion with service context, taking in multiple monitoring perspectives – the service-centric sum of the component parts.

Delivering Service Observability

Unlike other solutions Interlink places observability outputs in a business service context, aligned to service level objectives (SLOs) and key user journeys.

The ability to draw meaning from the visibility that individual monitoring tools bring and more closely align the technical monitoring output with what these mean for the health of your business services, service level objectives (SLOs) and key user journeys. Interlink’s AIOps platform delivers the following:

1. Based on business priorities, rather than just IT priorities
2. Data-driven service models and associated meta-data
3. Visualizations of service health across tools and domains
4. A foundation for orchestration, management and action

Realizing the value of data sources that can plug blind spots, we bring in normalized and enriched feeds from ITSM, including incident, change and CMDB information which is then correlated to service models.

Data-driven service models allow users to orchestrate, manage and act based on business priority, rather than IT priority - implementing the crucial missing link between IT and the business.

Service modelling constitutes relationship-driven observability, helping to pinpoint the root cause of issues, reveal what’s changed in your environment, who needs to be notified about it and what action is required to protect service availability.

Delivering a unified, single source of truth, accelerating detection of issues and incident response

1. Apply machine learning: aggregate, correlate, cluster and prioritize event data
2. Visualize/understand dependencies between components
3. Understand what is slow or broken, no matter when or where it happens and quickly understand exactly what needs to be done, slash MTTR, reduce outages
4. The ability to infer and predict the health of a complex system, explore properties and patterns not defined in advance – in a single pane
5. Correlate and prioritize event and incident data, automatically detect unusual changes, discover unknown-unknowns impossible to monitor by humans alone.