Five Logs that Could Reveal a Data Breach in your SAP Systems

One of the most important discoveries uncovered by security researchers investigating the recent data breach at Anthem is that the original compromise may have occurred as early as April 2014, nine months before the breach was discovered by the organisation.  The attack has led to the loss of personal information impacting over 80 million individuals. The investigation into the impact on health records stored by the organisation is ongoing. Such records have a far higher value in underground markets than financial data including banking and credit card information.

Anthem was alerted of the breach after a system administrator learned that his logon credentials had been compromised and used by attackers to access servers containing sensitive data. The fact that the discovery was made by Anthem itself should be applauded. The majority of breaches are not. Most are detected by law enforcement agencies, third parties, and even customers. However, the time lag between the initial breach and its eventual discovery is a concern and one that is consistent with most other successful attacks. According to the 2014 Verizon Data Breach Investigations Report (DBIR) based on an analysis of 1300 confirmed data breaches and 63,000 security incidents, the gap between the average time taken by attackers to compromise their targets and the time taken by victims to discover a breach has been widening steadily since 2004. This suggests that attackers are developing and exploiting ever more effective methods to compromise organisations at a rate that outstrips the ability of companies to detect and defend against such attacks. This is despite higher spending on both security solutions and personnel.

Protecting information in SAP systems from attack vectors used successfully against organisations such as Anthem requires two critical countermeasures. The first is system hardening. The second is log monitoring. This article focuses on the second of these measures. The effective and timely review of forensic data captured by several SAP logs can enable your organisation to drive back attacks before they lead to a data breach.

The first category of logs covers network traffic patterns. Incoming and outgoing connections registered in ICM/ Web Dispatcher, SAProuter, message server and gateway server logs should be regularly reviewed for suspicious network activity. This includes connection attempts from unknown or unauthorized source IPs or during unusual hours, as well as sessions that involve the transfer of large volumes of bytes to external destinations. The latter is a clear sign of potential data theft.

The second category covers authentication and authorization logs that record logon attempts and the actual resources accessed after successful logons. The main source of such data in SAP systems is the Security Audit Log. However, for more granular information, you should review log entries in the Read Access Log which register views and changes to sensitive data fields. UME log events in the J2EE Engine can be monitored using the NetWeaver Administrator. Within this category, logon attempts using default accounts across multiple systems and during irregular hours are especially suspicious.

The third category covers changes for configuration settings, files, user accounts, documents, programs and tables.  Logging such changes will support the reconstruction of events and help contain any breach. Authorization, password and other changes impacting user master records are automatically stored in non-transparent SAP tables which can be viewed using transaction SU01. Change documents can be used to capture changes to sensitive data objects. Changes to critical tables can be logged using SE13 and analyzed through report RSTBHIST. Changes to productive systems implemented through SAP transports are recorded in CTS and TMS logs stored in both transport directories and tables E070 and E071. Changes to profile parameters in managed systems, including security-relevant areas, are logged in Solution Manager and can be analyzed using Configuration Validation or Change Analysis.

The fourth category covers application and system events that are not directly security-relevant but may indicate potential malicious activity. This includes system shutdowns and restarts, unscheduled or unauthorized backups and error messages for the usage of memory, disk, CPU and other system resources. Such information can be collected from Syslog and other host-level event logs. It can also be accessed through local or central SAP System logs using transaction SM21.

The final area covers database-level actions and events, particularly activities performed by privileged non-system users including the execution of ALTER, INSERT and DELETE commands and CREATE and GRANT schema changes. You can minimize the performance impact of database logging in some database versions and releases by creating context-dependant policies that limit logging to precise scenarios. Examples include database connections originating during specific time periods or from outside specific application servers identified by hostname or IP address.

Attackers may attempt to remove evidence of their actions by altering or deleting log records. Therefore, it is important to secure access to SAP tables and OS-level files containing log information. Also, log files should be replicated to independent time-synchronized servers and log data held directly in SAP systems should be periodically archived using the archiving transaction SARA.