An innovative high performance cloud computing service

An innovative high performance cloud computing service

VSC – Virtual Supercomputer – is an innovative high-performance cloud computing service. When a system’s full computing power is not being used by its owners, the power that is available can be provided to clients.

VSC provides a ready-made, fully-functional analogue of a physical supercomputer that is comprehensively equipped with software and requires no specific knowledge of software installation or setup procedures. This allows owners of supercomputer facilities and independent software vendors to reach customers that have never before had access to supercomputing power.

VSC uses Clustrx and Xpandrx, operating systems for HPC (high-performance computing), cloud computing, and data centres. Unlike other HPC stacks, neither Clustrx nor Xpandrx is just a set of utilities, but a tightly integrated solution that binds together a set of scalable services to achieve un- precedented scalability and reliability.

Clustrx is used on clusters in Moscow University, including Lomonosov, no. 13 in the current TOP500 Supercomputing Sites. Xpandrx is the next evolutionary step in the Clustrx line, and is becoming an unbeatable platform for private and public clouds.

Clustrx/Xpandrx has been developed as a platform for future exascale supercomputers. A low-level hypervisor runs mobile virtualised instances of OS- level services on physical servers and makes up a secure distributed interconnected layer. Every service can work within multiple virtual machines that run on multiple physical servers in a management facility. Each piece of hardware and software in the whole cluster or cloud are under control of a real- time monitoring subsystem. The granularity and integrity of the monitoring data enables the delivery of a complex computing functionality.

VSC instances use a lightweight virtualisation scheme that provides a secure, isolated network environment to each virtual cluster, without de- graded performance. VSC instances that are ready for use will be available immediately through an online shop, which will be a marketplace for infra- structure providers to offer computing power and for independent software vendors to offer their software packages.

Why Erlang

  • Multiple flows of data being processed in parallel and redirected on-the-run seamlessly fits into the Erlang methodology.
  • Erlang’s fault tolerance model fits to the reliability needs set by most modern system software and enables implementation of trouble-proof distributed systems with only minimum effort.
  • The live code upgrade feature in Erlang simplifies implementation of non-stop system services.

 

Customer benefits

  • Users have easy, on-demand access to computing resources with high performance or specific configurations, with no need to maintain their own facilities.
  • Owners of supercomputing facilities and data centres can monetise surplus computing power and create a market for distribution of computing power/services.
  • Vendors can vastly expand the list of specialised software products available on their platforms and find a wider distribution market with a new category of users.

Product Data

  • Developed by:
    Massive Solutions Ltd
  • Target industries:
    • High-tech and engineering companies with heavy computing needs
    • FInancial or business analysis, data mining
    • Independent software developers and vendors
    • Supercomputing facilities

Technical Data

  • Programming languages:
    • Erlang
      parallel processing, resource distribution, system services
    • C
      core functionality, kernel modifications
    • Java
      portable interfaces
  • Development environment:
    • Linux (code)
    • Clustrx (testing)