Cloud computing systems
A multi-tier cloud-metering framework for metering cloud resources is presented. The proposed framework is capable of relating metering indicator of distributed resources contributing to the same application and located at different architectural levels. Metering data is collected from the front-end, the cloud nodes and VMs, and propagated through the middle-tier of the metering architecture to be processed and stored at the back-end. The design of the metering framework adopts a service-oriented architecture which allow the processing of cloud metering objects CMOs using service processing over open web transport protocol. To be able to build the middle and back-end tier, a SOA based on IPIoE and BOSML is presented which allow full consolidation of cloud resources achieving scalability and reducing the amount of wasted resources due to cloud resources de-fragmentation.
1. A system comprising:
a cloud computing system comprising a plurality of processors and memory operatively connected by a data communications network for executing a plurality of cloud computing applications for providing software services, wherein the cloud computing system comprises:
a first physical cloud node comprising a first local processor core local to the first physical cloud node; and
a second physical cloud node comprising a second local processor core local to the second physical cloud node;
a cross hypervisor, implemented on the cloud computing system, configured for building a virtual machine that spans across at least one physical boundary between the first physical cloud node and the second physical cloud node over the data communications network by starting a respective hypervisor on each of the first physical cloud node and the second physical cloud node,
wherein each hypervisor is configured to execute an operating system that adopts a service oriented architecture, and wherein each hypervisor is configured to support inter-processor interrupts issued by the first local processor core and the second local processor core received from the cross hypervisor over the data communications network and across the physical boundary between the first physical cloud node and the second physical cloud node, thereby causing the first local processor core to respond to at least a first inter-processor interrupt issued by the second local processor core as a valid inter-processor interrupt even though the first inter-processor interrupt was received over the data communications network; and
wherein each hypervisor is configured to adopt the service oriented architecture by virtue of a bare metal operating system markup language (BOSML) and a BOSML interpreter configured for executing one or more BOSML scripts to run on bare metal, and wherein each hypervisor is configured to support inter-processor interrupts received from the cross hypervisor using the BOSML interpreter.
2. The system of claim 1 , wherein the BOSML interpreter is configured for executing the one or more BOSML scripts to run on bare metal without any dependencies to intermediate dynamically linked libraries.
3. The system of claim 1 , wherein each hypervisor is configured to support inter-processor interrupts by virtue of exchanging inter-processor interrupt over Ethernet (IPIoE) protocol messages based on raw Ethernet packets.
4. The system of claim 1 , wherein the cross hypervisor is configured for presenting a global view of the two physical cloud nodes as a unified resource pool.
5. The system of claim 1 , wherein the cross hypervisor is implemented as a virtual hypervisor constructed over a set of distributed physical hypervisors.
6. The system of claim 5 , wherein each distributed physical hypervisor is configured to execute a BOSML-enabled operating system.
7. The system of claim 6 , wherein each distributed physical hypervisor is configured, by virtue of including a network stack in the BOSML-enabled operation system, to support inter-processor interrupts by virtue of exchanging Ethernet messages over the data communications network and across the physical boundary between the first physical cloud node and the second physical cloud node.
8. The system of claim 5 , wherein the cross hypervisor is implemented by extending one of the distributed physical hypervisors to support inter-processor interrupts by virtue of exchanging Ethernet messages over the data communications network and across the physical boundary between the first physical cloud node and the second physical cloud node.
9. The system of claim 8 , wherein extending one of the distributed physical hypervisors comprises supporting an inter-processor interrupt over Ethernet (IPIoE) protocol messages based on raw Ethernet packets.
10. A method comprising:
executing a plurality of cloud computing applications for providing software services on a cloud computing system comprising a plurality of processors and memory operatively connected by a data communications network, wherein the cloud computing system comprises:
a first physical cloud node comprising a first local processor core local to the first physical cloud node; and
a second physical cloud node comprising a second local processor core local to the second physical cloud node;
implementing, on the cloud computing system, a cross hypervisor configured for building a virtual machine that spans across at least one physical boundary between the first physical cloud node and the second physical cloud node over the data communications network by starting a respective hypervisor on each of the first physical cloud node and the second physical cloud node,
wherein each hypervisor is configured to execute an operating system that adopts a service oriented architecture, and wherein each hypervisor is configured to support inter-processor interrupts issued by the first local processor core and the second local processor core received from the cross hypervisor over the data communications network and across the physical boundary between the first physical cloud node and the second physical cloud node, thereby causing the first local processor core to respond to at least a first inter-processor interrupt issued by the second local processor core as a valid inter-processor interrupt even though the first inter-processor interrupt was received over the data communications network; and
wherein each hypervisor is configured to adopt the service oriented architecture by virtue of a bare metal operating system markup language (BOSML) and a BOSML interpreter configured for executing one or more BOSML scripts to run on bare metal, and wherein each hypervisor is configured to support inter-processor interrupts received from the cross hypervisor using the BOSML interpreter.
11. The method of claim 10 , wherein the BOSML interpreter is configured for executing the one or more BOSML scripts to run on bare metal without any dependencies to intermediate dynamically linked libraries.
12. The method of claim 10 , wherein each hypervisor is configured to support inter-processor interrupts by virtue of exchanging inter-processor interrupt over Ethernet (IPIoE) protocol messages based on raw Ethernet packets.
13. The method of claim 10 , wherein the cross hypervisor is configured for presenting a global view of the two physical cloud nodes as a unified resource pool.
14. The method of claim 10 , wherein the cross hypervisor is implemented as a virtual hypervisor constructed over a set of distributed physical hypervisors.
15. The method of claim 14 , wherein each distributed physical hypervisor is configured to execute a BOSML-enabled operating system.
16. The method of claim 15 , wherein each distributed physical hypervisor is configured, by virtue of including a network stack in the BOSML-enabled operation system, to support inter-processor interrupts by virtue of exchanging Ethernet messages over the data communications network and across the physical boundary between the first physical cloud node and the second physical cloud node.
17. The method of claim 14 , wherein the cross hypervisor is implemented by extending one of the distributed physical hypervisors to support inter-processor interrupts by virtue of exchanging Ethernet messages over the data communications network and across the physical boundary between the first physical cloud node and the second physical cloud node.
18. The method of claim 17 , wherein extending one of the distributed physical hypervisors comprises supporting an inter-processor interrupt over Ethernet (IPIoE) protocol messages based on raw Ethernet packets.