White House e-mail forensics case won't be easy to crack
The ongoing drama of missing White House e-mail messages — however it finally ends — highlights the complexities and costs any government agency could face if suddenly asked to account for e-mail or other networked data.
Finding messages that have been stored properly can be difficult enough.
Recovering messages by parsing, analyzing and sometimes combining data from hard drives and other media — the art of computer forensics — is considerably more difficult.
White House officials face a possible court order that could require them to create forensic copies of e-mail server backup tapes and, possibly, all computers and workstations that might contain e-mail messages dating to March 2003.
The order could affect a large number of systems in addition to e-mail servers.
In testimony before the House Oversight and Government Reform Committee in February, Theresa Payton, chief information officer at the Executive Office of the President’s Office of Administration, said her organization manages sensitive-but-unclassified e-mail accounts for more than 3,000 employees.
Forensic copies are not simply duplicates of hard drives and tapes. They are bit-by-bit copies that must be verified as being exact matches of the originals, said Benjamin Carmitchel, president and chief executive officer at ESS Data Recovery.
Given the time that has elapsed since the e-mail messages allegedly disappeared, it would be “extremely complex, if not impossible” to get verified copies of e-mail messages from that two and a half-year period, Carmitchel said.
Even if that were possible, it would be expensive. The forensics recovery process requires building an index and creating an analysis program to search the forensic copies. For each machine, that process would require several hours and cost as much as $1,500, said Richard Smith, a forensic specialist at Boston Software Forensics.
A similar forensics analysis would be significantly more complex and expensive for larger agencies.
Large organizations typically have 10,000 to 50,000 computers spread across a wide geographical area, said Eric Robi, president of California-based Federal Forensics Group.
Even if you could access those machines with forensics software, bandwidth limitations would create challenges, he added.
Much of the work can be automated, but some level of manual processing is always involved, Robi said.
“It can take months and require significant man-hours and other resources,” he said. “It’s a large-scale acquisition.”
Brian Robinson is a freelance writer based in Portland, Ore.