Short Bytes: Six MIT researchers are working to make a file system that will never lose data in case of a crash event. This could lead to the advent of crash tolerant computers and applications in many other domains.
Six MIT research boffins have demonstrated a file system capable of recovering all data in the event of a computer crash. This team will present this file system at the ACM Symposium on Operating Systems Principles in October. According the researchers, this system is mathematically guaranteed not to lose track of data during crash events.
Nickolai Zeldovish, an associated professor of computer science and a part of the team says, “What many people worry about is building these file systems to be reliable, both when they’re operating normally but also in the case of crashes, power failure, software bugs, hardware errors, what have you.”
Right now this file system is slow as compared to the current standards and the team is working to develop a reliable and efficient file system.
He explains that making sure that a file system can recover from a crash is tricky as there are so many ways your file system could crash. Thus, every disk operation and instruction must be considered.
To make their file system and studies different from the previous ones, the team has constructed a proof for the infallible file system. After this, they created the actual code and tested it using formal verification. For those who don’t know, formal verification is a step that determines the acceptable bound of operations for a system and involves mathematically proving that they’ll never exceed.
The team has checked their work against the actual file system and these proofs indicate that this file system is going to come into existence and it will be applied in many other domains apart from the crash tolerant computers.
Did you like the idea of file systems that will make crash tolerant computers? Tell us in the comments below.