Skip to main content

Using the Security Event Graph to Drive Alert Prioritization

One of the biggest differentiators at Sift Security is our security event graph: We map security events into a graph database. We then analyze the graph structure to prioritize alerts. Specifically, we look for clusters of interrelated alerts, score the clusters, and surface the clusters to the analyst. The analyst can then investigate each cluster in order, quickly assessing the threat and resolving the alerts in bulk.  

Our algorithms do the important work of sifting through isolated alerts and separating the false alarms and low priority alerts from high priority security incidents. We identify the high priority incidents by analyzing how alerts are related to each other. Key to this approach is our security event graph. This graph is stored in a graph database, a relationship-centric database that enables rapid execution of complex queries that would be very expensive to make in a traditional RDBMS.  The graph structure enables us to rapidly traverse relationships and find interrelated alerts. We use JanusGraph, an open-source, scalable, and distributed graph database, to store, query, and analyze our security event graph. 

The clusters are scored and ranked based on a number of factors, including:
  • Number and priority of alerts
  • Number of alert sources and types
  • Number of different affected entity types
In this post, we share two examples of how our graph algorithms provide effective alert prioritization, one from a traditional SIEM use case and another from our CloudHunter product.

SIEM Alert Prioritization

In this scenario, we have ingested Windows Event, netflow, Cylance, and Cisco IronPort logs. We have alerts from:
  • Threat intelligence sources
  • Cylance Antivirus
  • Sift Security’s Entity Behavioral Analytics module
  • Sift Security’s Detection Rules module
In total, there are over 1,000 alerts affecting some 70 host and 10 users.
The graph algorithms identify one cluster of alerts that stands out above the rest.  This cluster contains 55 alerts, 44 of which are high risk.  We display a summary of the alerts, including a breakdown of the alert priority scores by entity, shown below.

Screen Shot 2017-08-01 at 11.18.22 AM.png

We also surface an interactive visualization of this cluster of alerts to help the analyst quickly assess the threat.  The visualization (below) shows the user credential “Robert” at the center of the cluster, and 3 affected hosts: ws02, ws35, and dev.  The processes shared between ws35 and ws02 are a malware exploit identified by Cylance, the processes shared between ws02 and dev are reconnaissance processes identified by our Entity Behavioral Analytics, and the IP address is a threat intelligence hit.  Together, this cluster represents a successful malware exploit affecting 3 hosts and 2 user credentials.  
Screen Shot 2017-08-01 at 11.24.07 AM.png

CloudHunter Alert Prioritization

In the CloudHunter example, we have ingested CloudTrail logs from multiple AWS accounts.  The alerts come from two places:
  • Sift Security’s Analytics Module
  • Sift Security’s Detection Rules Module
Because this is a CloudHunter deployment, the detection rules module is preconfigured to alert on the Center for Internet Security benchmarks, and the analytics are configured for common cloud security use cases, such as identifying privilege escalation, data exfiltration, and other misuse of resources.
From thousands of individual alerts, the graph algorithms again identify a high priority cluster (below).  This cluster shows demo-contractor creating an instance from an AMI, and automated-s3-user accessing sensitive documents in s3 from that instance.  This is a situation where an insider was able to get access to data that should not have been able to by abusing the privileges they were granted. 

Screen Shot 2017-08-01 at 11.54.11 AM.png


The benefit of this graph-based alert prioritization approach is that these clusters of alerts may have otherwise gone unnoticed for a while, buried in a pile of other similar alerts. Our graph based alert prioritization helps reduce the mean time to remediation by enabling analysts to be more effective in their investigations.  If you’d like to try out our graph based alert prioritization techniques for yourself, consider trying CloudHunter, which can be deployed in under and hour to start identifying prioritized security incidents in your AWS cloud infrastructure.

For other posts about alert prioritization, see:

Popular posts from this blog

Data Exfiltration from AWS S3 Buckets

You will have no doubt heard by now about the recent Booz Allen Hamilton breach that took place on Amazon Web Services – in short, a shocking collection of 60,000 government sensitive files were left on a public S3 bucket (file storage in Amazon Web Services) for all to see. We are all probably too overwhelmed to care, given all the recent breaches we have been hearing about in the news. But with this breach it was different, it involved a trusted and appointed contractor whose job it was to follow security policies, put in place to avoid such incidents. So was this incident accidental or malicious? More, later about the tools we can use to tell the difference between the two. First, lets recap what happened. The Incident According to Gizmodo , the 28GB of data that was leaked not only contained sensitive information on recent government projects, but at least a half dozen unencrypted passwords belonging to government contractors with Top Secret Clearance – meaning anyone who got

Sift Joins Netskope, the Cloud Security Leader

Four years ago, we started Sift with the mission of simplifying security operations and incident response for the public cloud. In that time, we have assembled a fantastic team, created an innovative cloud detection and response solution, and have worked with many market-leading customers. I’m delighted to share that we’ve taken yet another step forward — as announced today, Sift is now officially part of Netskope. You can read more about this on Netskope CEO Sanjay Beri’s  blog  or in the official  announcement  on the Netskope website. For our customers, investors, partners, and team, this is an exciting new chapter. Let me tell you why we’re so excited.  Since the beginning, Netskope has had an unmatched vision for the cloud security market. Having started in 2012, they initially focused on SaaS security and quickly followed that with IaaS security capabilities. Six years later, they are now more than 500 employees strong and used by a quarter of the Fortune 100. They are a l

Sift Security vs. Elastic Search and Elastic Graph

We are often asked, “What is the difference between Sift Security and Elastic Graph ?” This is a great question that typically comes from folks who are already familiar with Elasticsearch [0] and Elastic Graph [1]. The answer boils down to the following: Elastic Graph is a tool for visualizing arbitrary aggregate search results. Elasticsearch is a Restful search that distributed, and has analytics engine that solves a number of use cases such as mapping from Python to ES REST endpoints. Sift Security uses a graph database to simplify and accelerate specific security use cases. In this blog post, we describe the advantages of each of these approaches, and conclude with a discussion of when to use each. Advantages of Sift Security vs ElasticSearch and Elastic Graph Query speed Sift Security builds a property graph to represent security log events at ingestion time.  We do this work at ingestion time for one reason:  to speed up common investigative queries.  When investi