Azure Security Center helps ESHA prevent, detect, and respond to threats with increased visibility into (and control over) the security of our customers Azure resources. It provides integrated security monitoring and policy management across our subscriptions. It helps detect threats that might otherwise go unnoticed, and it works with a broad ecosystem of security solutions.
Azure Monitor
Azure Monitor is basic tool for monitoring services running on Azure. It gives ESHA infrastructure-level data about the throughput of a service and the surrounding environment.
Azure Monitor includes the following components.
Azure Activity Log
The Azure Activity Log provides insight into the operations that were performed on resources in our subscription.
Azure diagnostic logs
Azure diagnostic logs are emitted by a resource and provide rich, frequent data about the operation of that resource. The content of these logs varies by resource type.
Windows event system logs are one category of diagnostic logs for VMs. Blob, table, and queue logs are categories of diagnostic logs for storage accounts.
Diagnostic logs differ from the Activity Log. The Activity log provides insight into the operations that were performed on resources in our subscription. Diagnostic logs provide insight into operations that our resource performed itself.
Metrics
Azure Monitor provides telemetry that gives ESHA visibility into the performance and health of our workloads on Azure. The most important type of Azure telemetry data is the metrics (also called performance counters) emitted by most Azure resources. Azure Monitor provides several ways to configure and consume these metrics for monitoring and troubleshooting.
Azure Diagnostics
Azure Diagnostics enables the collection of diagnostic data on a deployed application.
Threat Intelligence
Microsoft has an immense amount of global threat intelligence. Telemetry flows in from multiple sources, such as Azure, Office 365, Microsoft CRM online, Microsoft Dynamics AX, outlook.com, MSN.com, the Microsoft Digital Crimes Unit (DCU), and Microsoft Security Response Center (MSRC).
Researchers also receive threat intelligence information that is shared among major cloud service providers and subscribes to threat intelligence feeds from third parties. Azure Security Center can use this information to alert you to threats from known bad actors. Some examples include:
Harnessing the Power of Machine Learning - Azure Security Center has access to a vast amount of data about cloud network activity, which can be used to detect threats targeting your Azure deployments. For example:
Brute Force Detections - Machine learning is used to create a historical pattern of remote access attempts, which allows it to detect brute force attacks against SSH, RDP, and SQL ports.
Outbound DDoS and Botnet Detection - A common objective of attacks targeting cloud resources is to use the compute power of these resources to execute other attacks.
New Behavioral Analytics Servers and VMs - Once a server or virtual machine is compromised, attackers employ a wide variety of techniques to execute malicious code on that system while avoiding detection, ensuring persistence, and obviating security controls.
Azure SQL Database Threat Detection - Threat Detection for Azure SQL Database, which identifies anomalous database activities indicating unusual and potentially harmful attempts to access or exploit databases.
Behavioral analytics
Behavioral analytics is a technique that analyzes and compares data to a collection of known patterns. However, these patterns are not simple signatures. They are determined through complex machine learning algorithms that are applied to massive datasets.
They are also determined through careful analysis of malicious behaviors by expert analysts. Azure Security Center can use behavioral analytics to identify compromised resources based on analysis of virtual machine logs, virtual network device logs, fabric logs, crash dumps, and other sources.
In addition, there is correlation with other signals to check for supporting evidence of a widespread campaign. This correlation helps to identify events that are consistent with established indicators of compromise.
Some examples include:
Suspicious process execution: Attackers employ several techniques to execute malicious software without detection. For example, an attacker might give malware the same names as legitimate system files but place these files in an alternate location, use a name that is very like a benign file, or mask the file’s true extension. Security Center models processes behaviors and monitors process executions to detect outliers such as these.
Hidden malware and exploitation attempts: Sophisticated malware can evade traditional antimalware products by either never writing to disk or encrypting software components stored on disk. However, such malware can be detected using memory analysis, as the malware must leave traces in memory to function. When software crashes, a crash dump captures a portion of memory at the time of the crash. By analyzing the memory in the crash dump, Azure Security Center can detect techniques used to exploit vulnerabilities in software, access confidential data, and surreptitiously persist within a compromised machine without impacting the performance of your machine.
Lateral movement and internal reconnaissance: To persist in a compromised network and locate/harvest valuable data, attackers often attempt to move laterally from the compromised machine to others within the same network. Security Center monitors process and login activities to discover attempts to expand an attacker’s foothold within the network, such as remote command execution, network probing, and account enumeration.
Malicious PowerShell Scripts: PowerShell can be used by attackers to execute malicious code on target virtual machines for a various purposes. Security Center inspects PowerShell activity for evidence of suspicious activity.
Outgoing attacks: Attackers often target cloud resources with the goal of using those resources to mount additional attacks. Compromised virtual machines, for example, might be used to launch brute force attacks against other virtual machines, send SPAM, or scan open ports and other devices on the Internet. By applying machine learning to network traffic, Security Center can detect when outbound network communications exceed the norm. When SPAM, Security Center also correlates unusual email traffic with intelligence from Office 365 to determine whether the mail is likely nefarious or the result of a legitimate email campaign.
Anomaly Detection
Azure Security Center also uses anomaly detection to identify threats. In contrast to behavioral analytics (which depends on known patterns derived from large data sets), anomaly detection is more “personalized” and focuses on baselines that are specific to your deployments. Machine learning is applied to determine normal activity for your deployments and then rules are generated to define outlier conditions that could represent a security event. Here’s an example:
- Inbound RDP/SSH brute force attacks: Your deployments may have busy virtual machines with many logins each day and other virtual machines that have few or any logins. Azure Security Center can determine baseline login activity for these virtual machines and use machine learning to define around the normal login activities. If there is any discrepancy with the baseline defined for login related characteristics, then an alert may be generated. Again, machine learning determines what is significant.