Live Forensics and the Cloud Part I

Within the realms of digital forensics analysts traditionally performed analysis on static data, either from a core dump, bit to bit imaging etc. Recently we have seen an increased focus directed at the live forensics environment. As users rely more on mobile and other remote devices to access data on demand; data possibly held in some manner of cloud environment, investigators will have to adapt their mode of investigations to suit.

I recall reading a marketing pitch a while aback where some vendor claimed that an advantage of Cloud Computing is, an ability to conduct live forensics without disrupting mission critical systems. How effective this claim may be, is subject to examination.

According to Brian Carrier - "The only difference between a live and a dead analysis is the reliability of the results; a live analysis techniques use software that existed on the system during the time-frame being investigated; dead analysis techniques, use no software that existed on the system during that time-frame." - Bear in mind though that there are different aspects and levels to these statements.

A few of the experts in this field with whom I was able to interact whilst conducting graduate research a while aback, did state that when conducting a live analysis, the system under investigation will inevitably be altered in some manner or another.This in essence can define a live analysis as not being a pure forensic form.

However the potential for gaining valuable data is looked on as the lesser of two worse case scenarios in this instance.

As we know the concept of cloud computing is an amalgamation of already existing sundry computing concepts viz. distributed, grid and utility computing.

The Cloud Computing environment is as we know susceptible to classical attacks (Cross Site Scripting,DDoS,etc ) as is any regular system. A concern for any security consultant can be the potential for exploitation of a system under live analysis.

Thus the health of any one cloud ecosystem lies within the domains which ensure that confidentiality and privacy concerns within cloud computing are effectively monitored, managed and mitigated. This will include the area of digital forensics and its place within the e-discovery process.

In regular systems one instance of exploitation can be with rootkits. Rootkits as we know can be divided into database rootkits and BIOS rootkits. The potential for exploiting both and remain undetected is high e.g.by manipulating the ACPI (Advanced Configuration Power Interface) in BIOS via its ASL programming language to modify hardware features or memory.

Hypothetically speaking one may be able to insert a rootkit which reacts to a forensic probe and then output pre-programmed results to suit an attacker; remember that a snapshot of a running system can be only reproduced up to its specific instance and cannot be reproduced at a later time-frame. As a result with a live system, data from a probe up to one instance will be different from data from another probe say 15 minutes into the live system.

In another scenario a rootkit may be programmed to respond to a probe by purging and shutting down the system - According to R.C. Vernon "A "hard" reboot includes a power cycle, which ensures that sensitive information in volatile memory is purged".

One can then question; what happens if an attack is able to compromise a host's cloud system and insert a rootkit which remains undetected as described above? Can multiple tenants of any one cloud ecosystem be compromised and if so how far can any such exploitation propagate without detection?

Within the cloud users expect their identity and data to remain private via anonymous authentication; if per chance a system is compromised, an attacker then take advantage of this factor of anonymous authentication and possibly spoof any tenant within the cloud as the attacker while data is compromised. How can an investigator identify and track such an issue?