ALERTS THAT MAKE A DIFFERENCE
What makes us special?
Paths Analysis
Goal: Path travelsal is very importatn in indentification of related issues and understanding of whole issue process. It is equivalent to performing complicated joins. Here with graph structure it is natural investigation step.
Example: One intuite example is identification of impacted microservices. You can travelsar through topology of services startting from some alert's symptom and ending on overloaded disc space of some DB.
Analysis of problem's borders and responsibilities
Goal: Problems are often not fully identified, partial identification causes an incorrect estimation of the actual risk with inproper actions.
Example: Thanks to the multimodal knowledge graph, which also encodes information about business and technological boundaries, we are able to estimate the risk much more precisely and address the problem to the groups actually responsible for solving it.
Multimodal Correlations
Goal: Ability to correlate multimodal signals is one of main goal of Beemon. Correaltion of time series anomalies with rows of log file and printscreen of system's page is what SecOps really need.
Example: Metric of increased network latency correlated with few servers ping failure and errors in those servers logs is what really presents a situation under one single incident. This is how we try to cluster alerts, metrics and all other multimodal data.
Clustering and classification of similar alerts
Goal: Elimination of redundancy of information about the same problem.
Example: In the IT world, often several applications depend on one specific service. When such a hub service stops working properly, it causes an avalanche of alerts, which should be clustered into one problem, and then the notification addressing it should go to the appropriate team.
Anomalies detection
Goal: Identification of security issues and operational problems is very hard. Levrage of structure of multimodal knowledge graph opens new viewpoint.
Example: Local structure of graph is often abnormal when system is crashing or some unusual activitieas are perform on the system. This non normal behavour of the system and other related system is easy to recognize on multimodal knowladge graph.
Get to know BEEMON!
Observability and security driven by multimodal knowledge graph
The central element of Beemon is the Multi-Modal Knowledge Graph (MMKG), which combines diverse data into a relational, structured form. The challenge we have solved was the ability to construct it online, in a dynamic and efficient performance way from data streams as inputs.
Contact form
Marketing department
Headquarters