Computers and password cracking are a natural fit.  From the Bletchley Park days when the first digital computers cracked Nazi Enigma codes, computers have been doing the hard work.  Whilst ordinary PCs now have a massive amount of computer power compared to old machines, the very fastest computers – supercomputers – are the first choice for fast recovery of lost passwords.

Read on to find out how supercomputers are used to crack passwords.

Firstly, what is a supercomputer?  In the past, it was something like a Cray I, with a few processor units, but hand built for maximum speed, or a top of the line Sun system with exotic CPUs.  Now, supercomputers are often based on commodity hardware and take advantage of massive parallel processing to get the job done.  They are primarily focussed on extremely fast processing.  Someone once asked me if a mainframe would deliver faster results than (say) a PC.  The answer generally is no.  A mainframe is designed for fast disk output, multitasking and printing, not massive CPU power.  Password recovery is pretty much a CPU bound problem, and that is why a supercomputer is needed.  

Finding passwords (or encryption keys) is inherently a parallel process – it is a search problem.  The work is to search through the possible passwords or encryption keys to find the one that fits.  The work can be divided into logical units and farmed out to separate processors, each of which run an identical program.  This means that the more processors you have, the faster you can get a result.

One approach is to build a cluster.   The military will cluster Cray systems together, but for our use, we need something cheaper.  For password cracking, this comprises a single control system, with multiple units that each take a part of the problem.  It is not necessary to have a Beowulf class cluster.  Each node shares a common logical drive and can read and write to it.  They load share by taking a part of the problem away, processing it, and then coming back to take the next part when it is finished.  This allows for a mixture of nodes – they don’t have to be identical in performance.  The central unit looks after giving each processor the next section of the search space.

Pages: 1 2

Tags: password recovery, supercomputer

This entry was posted on Saturday, December 27th, 2008 at 9:16 pm and is filed under Recovery Services. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.