MIT researchers eliminate data loss from computer crashes

Hardware errors, power failures, and software bugs won't matter anymore if some scientists' ideas are correct.

MIT researchers prevent data loss computer crashes
Credit: HGST

One of the issues with computer crashes is not so much that the machine has crashed—often a mere inconvenience—but that data was lost in the process.

While the computer is writing its ones-and-zeros, it loses track of what it's written and what it hasn't, and data becomes corrupted. It's been a problem since magnetic storage was invented.

However, MIT researchers think they've got a solution. They say that they've invented a file system that is guaranteed not to lose any data in a crash.

Mathematics

MIT's system uses mathematics to verify the data. It's based on a known technique called formal verification that in this case applies to the file system. The reliability of the file system is established through the formal verification process.

Formal verification is a way of proving or disproving correctness using mathematics.

Verification

"The acceptable bounds of operation for a computer program" are defined mathematically. Then the system proves that the "program will never exceed them," says Larry Hardesty of the MIT News Office, writing on its website.

The scientists say their system is slow, but that the concept behind the verification technique can be enhanced eventually, to make more sophisticated designs.

Crashing

"Making sure that the file system can recover from a crash at any point is tricky because there are so many different places that you could crash," Nickolai Zeldovich said on MIT's website. He is one of the three MIT computer-science professors on the new paper.

"You literally have to consider every instruction or every disk operation and think, 'Well, what if I crash now? What now? What now?'" he says.

Guaranteed no data loss

But the scientists do say that their formal verification technique is guaranteed not to lose data.

They say that their method proves "properties of the file system's final code, not a high-level schema," says Hardesty.

Therefore, it's better than anything else—although it is complicated and has been difficult to achieve.

Proofs

For one thing, they've had to develop what's called a "proof assistant," which provided a formal language for the system and relationships. Proofs are used around mathematics as a kind of sequence to verify things.

"Proofs are checked against the actual file system, not some whiteboard idealization that has no formal connection to the code," Adam Chlipala, another professor, says on the website.

Behaviors

Another complication that they had to deal with was describing the relationships "between the behaviors of these different components under crash conditions," Hardesty explains.

Determining that "the file system did, in fact, adhere to the logical relationships described in the proof," was another element to the work, Hardesty says.

'Crash-proof computer'

However, what they ended up with is the "world's first crash-proof computer," Wired says in an article about the technology.

It's a slightly misleading headline, in that "the computer system is not necessarily unable to crash, but the data contained within it cannot be lost," the Wired article's author correctly qualifies later in the story.

In any case, whatever you want to call it, guaranteed crash-tolerance is about to become a reality.

This article is published as part of the IDG Contributor Network. Want to Join?

To comment on this article and other Network World content, visit our Facebook page or our Twitter stream.
Must read: Hidden Cause of Slow Internet and how to fix it
Notice to our Readers
We're now using social media to take your comments and feedback. Learn more about this here.