Best Submit Onlywire!: Systems that never crash

Modern operating systems are based on principles that are more
than hundred years old. In simpler words, your programmes are
made up of a series of commands that are based on each other
and executed one after the other. Such a list-model can hardly
make full use of the capabilities of multi-core CPUs. It is equally
inefficient when it comes to dealing with malware; although
current operating systems have security mechanisms such as virus
scanners, they often only work once the system is already
damaged. Programmes with a built-in self-correction function
required a completely different concept. Researchers like Peter
Bentley from the University College London follow models that can
be drawn from nature. “Natural processes are decentralised and
fault-tolerant, and can heal themselves” and “a computer should
be able to do so too”, says Bentley.
Programmes that eliminate errors
The Systemic Computing (SC) principle dismisses the list-model.
So an SC programme can be made up of any number of systems,
which can be combined with each other. Every system contains
three components: two data inputs and a kernel that computes and
functions as data output. As a result, every system can be used as
a data input for another system. In this respect, an SC programme
is similar to the famous Russian Matryoshka dolls, which holds
several smaller copies of themselves. The systems are executed as
parallely as possible so that it never really comes to a complete
crash. Plus, there are several mechanisms in place in case the
systems make a mistake: identical copies that can be analysed and
rejected by a correction system; the area of application can also be
defined for these systems. To start with, Systemic Computing can
be used in niche applications, for example, those in traffic systems
or astronautics, where errors can be fatal.
The current systems cannot be relied on for not making such
mistakes. But even small, simple developments can be useful for
our computers. HP SureStart, a repair function for BIOS, is a
component that cements a huge gap in the current security
architecture. HP has recently started equipping its Business series
such as EliteBooks or ZBooks with this technology. HP notebooks
save a copy of the original BIOS and the Master Boot Record in
their own memory. If both central software components get
corrupted, they can be restored using these copies. Brendan Eich,
the chief technology officer at Mozilla is working on another
problem: he wants to implement a source check for the Open
Source browser Firefox. This check is supposed to exclude the
possibility of a manipulation of the source code, for example, by
intelligence services like NSA.
The US government too encourages the development of selfhealing software. The Defense Advanced Research Projects
Agency (DARPA) has also organised a Cyber Grand Challenge, a
national competition for developing self-healing systems that must
also be able to recover from malware attacks. In the first round,
participants are provided with a series of unsafe executable files
and the software is supposed to find and eliminate these errors.
Self-healing must be integrated in the final round in 2016.

Saturday 31 May 2014

Systems that never crash

More Resource

No comments:

Post a Comment