Zebrium RCaaS uses unsupervised machine learning on logs to automatically find the root cause of software problems. It does not require manual rules or training and typically achieves accuracy within 24 hours.
As logs are ingested, the ML analyzes them looking for abnormal log line clusters that resemble problems (i.e. abnormally correlated rare and error events from across all log streams). When it detects one of these “abnormal” clusters, it will generate a suggested alert.
A suggested alert contains metadata (e.g. title, description, etc.), a root cause report (set of correlated log lines that likely help to explain a problem) and a suggested alert rule (one or two log events types that form the signature for this type of alert). A user can choose to “accept” or “reject” a suggested alert. If a suggested alert is accepted, the user can edit the metadata and alert rule and can also decide upon the action to take if the same kind of alert type occurs again (e.g. to send a Slack, email or other notification).
If you currently use a monitoring tool from ScienceLogic, Datadog, New Relic, Elastic, Dynatrace or AppDynamics, we recommend installing one of our dashboard integrations. This will allow you to visualize suggested alerts on your existing monitoring dashboards.
To get started, you need to install one of our log collectors or upload a set of static log files. For instructions, click Log Collectors and File Upload in the menu on the left.
Questions? Contact: email to firstname.lastname@example.org