Anomaly detection tools are typically used to only track a limited set of KPIs for the following reasons:
- the user has to create the metrics (often this is dev work)
- handling of anomalies (typically only a graph showing the relevant timespan) require significant manual analysis
- aggregation, correlation between anomalies is manual
Loom seeks to save the user the need to prepare the metrics or investigate the anomalies, and relies on artificial-intelligence and machine-learning to automate these tasks. It constantly improves its detection quality by leveraging user feedback.
Furthermore, anomalies are not the only phenomenon that might interest a user. For example, we have found that the appearance of a new signal in the data is one of the best hints for a phenomenon that deserves attention, but this is invisible for anomaly detection tools. Loom does not take the mathematical-approach for anomaly detection, but instead aims to detect what is interesting for analysts – and often this goes beyond the definition of anomaly.
As for correlation between the anomalies, we developed advanced multi-layer entity based correlation engine that enable to reveal the real story behind each issue.
When Loom finds something interesting, it adds an actionableinsight in plain English, to help you resolve issues faster.
AIOps For Modern Hybrid IT
Watching Over Your Logs So You Don't Have To