Saturday, June 26, 2010

Another Reason for Detailed Access Logs

Another poorly written application, another data leak. Not new, barely news.

This statement is interesting though:

“[company spokesperson] said it's unclear how many customers' information was viewed, but that letters were sent to 230,000 Californians out of an "abundance of caution.”

Had there been sufficient logging built into the application, Anthem Blue Cross would have known the extent of the breach and (perhaps) could have avoided sending out all 230,000 breach notifications. That’s a view on logging that I’ve expressed to my co-workers many times. Logs can verify what didn’t happen as well as what did happen, and sometimes that’s exactly what you need.

There are a couple of other interesting things in the story:

“the confidential information was briefly accessed, primarily by attorneys seeking information for a class action lawsuit against the insurer.”

That’ll probably cost Anthem a bundle. Letting lawsuit-happy attorneys discover your incompetence isn’t going to be the cheapest way to detect bad applications.


“a third party vendor validated that all security measures were in place, when in fact they were not.”

Perhaps the third party vendor isn’t competent either?

Via Palisade Systems


  1. Logging isn't sufficient: you have to set up logging in such a way that you can trust it after a breach. Once a host is broken into, you can't trust anything on it, or anything coming from it.

    So if a machine does log remotely, and was breach at 08:59:59, then any log entries from 09:00 onwards can't be trusted.

    Setting up a properly auditable infrastructure is no easy thing if you've got a lot of moving parts.

  2. Good point.

    The default for me is that the only logs that I consider usable for forensics are those that are logged off the server, in real time, to a remote server on a separately firewalled network segment.

    The text that apps store on the server in the 'log' directory isn't a 'log' in the context of forensics.


  3. It's a struggle to get anyone to define requirements for logging that would be useful in a forensic analysis. For years, all I got in response to requests for such requirements -- from very well-meaning but very busy people -- was an acknowledgment that it's a good question and a general statement about logging any change in authentication/authorization status and any "substantive" change to data. That has begun to change.

    Clearly there are some applications, though, or at least parts of some applications, where viewing certain data should also be logged. I seriously doubt that this requirement would be identified without a risk analysis / threat model.