Obligatory disclaimer: I Am Not A Lawyer. “Evidentiary weight” is probably something that involves a set of scales.
Evidence of an electronic crime is most commonly retrieved through a well-established process of computer forensics. Regardless of the actual examination techniques used, the evidence (a seized hard drive, for example) has to be handled in such a way that its integrity is preserved, usually by way of examining a forensic duplicate of the media and leaving the original in the evidence safe.
It’s a little different when practicing network forensics:
- You have to be routinely collecting the evidence before the mischief is perpetrated. Network-based evidence is fleeting in nature – once it’s gone, it’s gone – if you’re not recording something all the time, when the mischief happens you’ll miss it.
- There isn’t such a firm concept of “original” or “best” evidence as there is in the computer forensics world. Packet captures are a copy of the original traffic, collected syslog messages are copies of log messages from a device, and netflow exports are descriptions of an act rather than the act itself. None of the copies are “forensically sound” like a disk image can be considered to be; they all sound rather like second-hand accounts.
A handy thing to do with all this evidence is to collect it, usually with some kind of SIEM or Log Management device, and vendors of such devices go to lengths to preserve the integrity of the collected data to maximise its weight as evidence. For example, a Cisco CS-MARS box stores everything in an internal Oracle database to which the user has no direct access; the idea is to give some assurance that the collected logs haven’t been tampered with. Other vendors like AlienVault say that their product “Forensically store(s) data (admissible in court)”. Given the potentially second-hand nature of our evidence, these anti-tamper measures are surely a good thing.
However, how can we convince someone that the evidence we are presenting is a true and accurate account of a given event, especially in the case where there is little or no evidence from other sources? Perhaps a laptop’s hard drive has been shredded, and the only evidence we’ve got is from our SIEM/LM box. Our evidence may be considered unreliable, since:
- It could be incomplete:
- A packet capture may not have captured everything – dropped packets are commonplace.
- UDP-based message transfer (syslog, netflow, SNMP trap, etc.) is best-effort delivery. Dropped messages will not be retransmitted, and most likely a gap in the sequence will not even be noticed.
- We may have exceeded our SIEM’s events per second threshold, and the box itself may have discarded crucial evidence.
- The body of evidence within our SIEM could have been tampered with.
But didn’t I say that vendors went to great lengths to prevent tampering? They do, but these measures only protect the information on the device already. What if I can contaminate the evidence before it’s under the SIEM’s protection?
The bulk of the information received by an SIEM box comes over UDP, so it’s reasonably easy to spoof a sender’s IP address; this is usually the sole means at the SIEM’s disposal to determine the origin of the message. Also, the messages themselves (syslog, SNMP trap, netflow, etc.) have very little provenance – there’s little or no sender authentication or integrity checking.
Both of these mean it’s comparatively straightforward for an attacker to send, for example, a syslog message that appears to have come from a legitimate server when it’s actually come from somewhere else.
In short, we can’t be certain where the messages came from or that their content is genuine.
So what could an attacker achieve by injecting false messages into an SIEM?
- They could plant false evidence of an act that did not take place. Perhaps we want to get someone fired for downloading questionable materials, for example.
- They could plant “tasty” evidence that the SIEM operators are likely to investigate first, buying the attacker time.
- They could force the box over its EPS limit so that genuine evidence is discarded.
- They could inject malformed messages that may lead the SIEM operators to believe that their box is somehow corrupted, and wait until it is offline for maintenance before striking.
- They could inject events or series of events that “couldn’t happen” to cause the SIEM operators to distrust the accuracy of the system (e.g. a proxy log message indicating a fetch of a webpage with no corresponding netflow export).
- Or perhaps their aim is to simply discredit the entire pool of evidence by injecting obviously false messages. Yes, an analyst could try to weed out the fake ones, but could they prove that they’d found all of them?
- <insert your own creative evil here>
If there’s any doubt as to the provenance of a message received by the SIEM, one could always go back to the original source and look for it there. The problem here is that it might not be there any more (due to log rotation on a server) or there may not even be an original to compare it to (a netflow export is the original message, for example).
I’m surely being a bit pessimistic here, but given that there’s a distinct possibility that any given SIEM is chock full of lies, how much evidentiary weight does its content actually possess? Does anyone know of any successfully-prosecuted cases where evidence from an SIEM has been key?
Alec Waters is responsible for all things security at Dataline Software, and can be emailed at alec.waters(at)dataline.co.uk