Optanix on AIOps – Part 2: The Capabilities of AIOps Tools in IT Operations Environments

In part one of our “Optanix on AIOps” series, we discussed the impact of AI for IT operations. In this installment, we’ll break down the capabilities of AIOps tools in IT operations environments.

How AIOps Tools Improve Data Ingestion & Handling

Consider the typical service provider ecosystem or large enterprise. There are many functional aspects to managing and monitoring the constantly expanding IT infrastructure that provides the backbone that service delivery depends upon. Most importantly, the infrastructure is dynamic – new services are provisioned, new technologies are incorporated, and network elements are added or upgraded on a persistent basis. 

AIOps Tools Normalize Fragmented Data Sets

As a result of this constantly morphing IT infrastructure environment, newer monitoring tools are routinely deployed on top of existing ones in an attempt to improve IT monitoring. This tool sprawl is a common side effect of growth, and it typically results in overlapping capabilities, such as multiple dashboards, control systems and analytics systems.

There are several categories of IT monitoring tools that contribute to tool sprawl. IT infrastructure monitoring (ITIM), network performance monitoring and diagnostics (NPMD), digital experience monitoring (DEM) and application performance monitoring (APM) tools are commonly deployed to meet domain-driven objectives.

But while these tools generate a vast amount of data, their data sets are typically fragmented, siloed and unexploited. This data is critical to providing valuable insights – but only if it can be consolidated effectively and used as part of a larger analytics strategy.

The challenge for AIOps tools is to ingest this data from across multiple sources and handle it in a manner that normalizes data formats and facilitates analysis that leads to actionable intelligence. The nature and source of the data is varied – today’s leading AIOps platforms capitalize on this diversity accordingly to generate value. The selection and filtering of collected data is key to the insights that follow. 

Analytical Outcomes

When ingesting data from diverse sources, simple raw aggregation is not enough – there must be data enrichment. With AIOps tools, such enrichment can occur in one of many ways and is tied to the ability to manage historical and real-time data streams:

• Historical data, such as logs and events, offers a retroactive view and can be enriched by applying metadata and tags to provide context for search indexing.
• Real-time data, such as performance and telemetry information, can be enriched to generate time-series data by overlaying data points with time stamps.

On top of enriching data, AIOps tools also add suitable labels that provide key-value pairs. This brings further value when historical and real-time information is accessed later on.

The next consideration is how the data is retained. To realize AIOps benefits, it is necessary to manage the consolidated data in either a data lake or some kind of data pipeline repository that provides centralized access and visibility. This centralized view makes it possible to conduct the kinds of analysis that ultimately deliver the advanced insights enabled by AIOps platforms.

AIOps tools provide the capability to aggregate events in an agnostic manner. Once data is ingested from multiple sources, enriched and stored in a consolidated fashion, a variety of analytical techniques can be applied to glean insights from it. These include:

• Historical analysis
• Anomaly detection
• Performance analysis
• Correlation and contextualization

The primary benefit of these outcomes is reduced MTTR. By collating data from across data streams and triaging it at the ingestion stage, it is possible to dramatically reduce duplication and volume of event data and bring about efficiencies in the signal-to-noise ratio.

This has a direct effect on the ability of an AIOps platform to provide superior causality outcomes. It also makes root cause determination faster by an order of magnitude.

AIOps for IT Operations Enables Actionable Decision-Making

AIOps platforms consolidate data from across monitoring and other ecosystems. They then apply advanced analytical techniques to generate insights that can then be acted upon. When it comes to determining causality, correlation and analysis, AI for IT operations further supercharges analysis by leveraging:

• Pattern recognition and machine learning capabilities that optimize correlation of events across various sources.
• Complex data streams that include text and metrics overlaid with time series and topology data, which enriches the raw information.

AIOps’ ability to analyze from end to end and top to bottom and to study real-time performance and metrics make it possible to detect abnormal conditions as they develop. Techniques such as the use of baselines, dynamic thresholds and peer comparison can all trigger alerts before an issue transpires that might actually impacts service delivery. This makes it possible to extrapolate events and prevent breakdowns, resulting in a predictive system.

On top of all this, AIOps tools enable automations that provide the ability to initiate action. This may be as simple as informing the appropriate resource with an alert, or as advanced as automatically kicking off predetermined actions from integrated automation libraries.

In our next installment in this series, we’ll take an inside look at specifically how Optanix’s AIOps technologies improve IT operations and support an agile IT infrastructure.