Free Market HPC Cloud In Development
Distributed computing has undergone many permutations, from its roots in grid computing to support large scientific endeavors to Sun-style utility computing, to the kind of public cloud computing popularized by Amazon Web Services. One of the newest projects of this type combines cycle sharing, think SETI@home, with cryptocurrency principles. Xennet, as it’s called, is being presented as a free-market alternative to Amazon Web Services.
Founder and innovator Ohad Asor describes the service as a distributed and decentralized arbitrary computation grid. The open market platform connects providers, those who make computation power available for a negotiated fee, with publishers, users who require compute cycles to run arbitrary computational tasks. Participants are free to pay or charge any rate they want. The setup relies on lightweight OS virtualization technology to ensure a safe computing experience.
“The network is 100% distributed and decentralized: there is no central entity of any kind, just like Bitcoin,” explains Asor. “All software will be open source. Publishers pay Providers directly, there is no middleman. Accordingly, there are no payable commissions, except for regular transaction fees which are being paid to XenCoin miners.”
The system’s building blocks consist of the lightweight OS virtualization tool, Docker, benchmarking technologies, an improved Blockchain technology, and a novel protocol. Software components include a Linux distribution (xennet-os) with Xennet code to manage Docker containers and collect measurements; a client software (xennetd and xennet-qt) that configures and manages the xennet-vm according to the preferences of publishers and providers; and blockchain (xencoind and xencoin-qt), implementing XenCoin as a cryptocurrency, being used as tokens to use computing machines.
Although XenCoin is not itself a currency, it can be monetized by using it to rent computational power used for mining cryptocurrencies (such as Bitcoin).
The upstart sees big potential in the big data space. Target applications include number-crunching, MapReduce, text analytics, predictive, or molecular dynamics tasks. “Typical Xennet publishers do not need one virtual machine – they need thousands of them,” notes company literature. “Take, for example, protein folding related computations. A single researcher wants to fold a certain protein. It would take him days or weeks to run computations on the university’s computer labs, if he gets lucky enough to get access to the university’s computation resources. He may get even luckier and have F@H fold it for him. But for most researchers, AWS may be the only practical option.”
Asor says that “presale” is expected and that “development is progressing at full pace.”