Virtual resource partitioning
 | This article has multiple issues. Please help improve it or discuss these issues on the talk page. (Learn how and when to remove these messages)
No issues specified. Please specify issues, or remove this template. |
Virtual resource partitioning (VRP) is an operating system-level virtualization technology that allocates computing resources (such as CPU & I/O) to transactions. Conventional virtualization technologies allocate resources on an operating system (Windows, Linux...) wide basis. VRP works 2 levels deeper by allowing regulation and control of the resources used by specific transactions within an application.[1]
In many computerized environment, a single user, application, or transaction can dominate all server resources and by that, affect the QoS & User experience of other active users, application or transactions. For instance: A single report in a Data warehouse environment can jam the Storage by producing high amount of IO/sec throughput. Or a CPU bound application may use all Server CPU power and cause CPU starvation to other activities. VRP technology allows balancing, regulating and manipulating the resource consumption of each and every transaction, and by that, improving the overall QoS, SLA & End user experience.
VRP technology overview
VRP is usually implemented at the Operating system in a way that is completely transparent to the application or transaction. The technology creates virtual resource lanes and redirects specific transactions to those lanes allowing them to take more or less resources according to the lane definition.
VRP technology can be implemented in any Operating system. Existing vendors in the market applied it on most of the Open operating system such as Windows, Red Hat, Suse, HP-UX, Solaris, tru64, AIX and others.
In any Operating system, the application is communicating with the Kernel in a different way which requires different implementation of VRP. Windows is managing resoruce allocation in a complete different way than Linux Kernel. Therefore applying virtual resoruce lanes implementation is usually kernel dependent. A safe implementation of VRP technology is usually made up of a mixture of several resource allocation techniques. VRP implementations depend on the transaction type, consumed resource or Kernel state. In a real and live system, Kernel state, transaction types and resource consumption are constantly changing and therefore the VRP technology should change the implemented technique accordingly in real-time.
References
- ^
Lipari, G.; Bini, E. (2–4 July 2003). "Resource partitioning among real-time applications". Real-Time Systems, 2003. Proceedings. 15th Euromicro Conference on. IEEE Xplore. pp. 151–158. ISBN 0-7695-1936-9.
{{cite conference}}: Cite has empty unknown parameter:|coauthors=(help); Unknown parameter|booktitle=ignored (|book-title=suggested) (help)CS1 maint: date format (link)