Virtualization system and resource allocation method thereof

This application claims the benefit under 35 U.S.C. §119(a) of a Korean patent application filed on Sep. 6, 2010 in the Korean Intellectual Property Office and assigned Serial No. 10-2010-0086968, the entire disclosure of which is hereby incorporated by reference.

1. Field of the Invention

The present invention relates to a virtualization system for supporting at least two operating systems and a resource allocation method of the virtualization system. More particularly, the present invention relates to a method for allocating hardware resources on multiple cores of the virtualization system supporting at least two operating systems and a system for executing the resource allocation method.

2. Description of the Related Art

FIG. 1 is a diagram illustrating a configuration of a virtualization system according to related art.

Referring to FIG. 1 , the virtualization system supporting at least two operating systems is capable of utilizing resources efficiently on multiple cores. An operating system running on the virtualization layer acts as if it occupies the corresponding hardware resource wholly on its own.

The virtualization software operates in a similar way as a scheduler of the operating system. Accordingly, the virtualization software schedules the individual operating systems running on the virtualization layer as tasks. The operating systems are assigned time resources in a predetermined ratio. In order for the at least two operating systems to share a hardware resource, each operating system use the time resources in up to a predetermined number of time slots (t). How to allocate time resources for each operating system is described below with reference to FIG. 2 .

FIG. 2 is a diagram illustrating a method for a plurality of operating systems to share a resource according to related art.

Referring to FIG. 2 , the first operating system (OS 1 ) 210 consumes the time resource allocated for tOS 1 and then hands over control to the Virtual Machine Monitor (VMM) as a hypervisor so as to execute tasks necessary for driving hardware. After completing the operations necessary for driving hardware, the VMM hands over the control to the second operating system (OS 2 ) 220 . The second operating system 220 consumes the time resource allocated for tOS 2 . The second operating system 220 then hands over the control to the VMM. A method for each operating system to operate with the allocated time resource is described below with reference to FIG. 3 .

FIG. 3 is a flowchart illustrating a method for an operating system to operate with an allocated time resource according to related art.

Referring to FIG. 3 , the VMM for driving at least two operating systems starts operating in step 310 . Here, the VMM may also be referred to as a hypervisor and control multiple operating systems running in a single host terminal.

The VMM creates a guest OS 1 as the first operating system and fixes the weight of the resource allocated to the OS 1 in step 320 . The VMM creates a guest OS 2 as the second operating system and fixes the weight of the resource allocated to the OS 2 in step 330 .

The VMM adjusts the time quantum per operating system in step 340 . The VMM determines whether the scheduler has scheduled the OS 1 or OS 2 in step 350 . If the OS 1 has been scheduled, the OS 1 runs in step 360 . If the OS 2 is scheduled, the OS 2 runs in step 370 .

In such a virtualization system, the time resource allocated according to the workload of the tasks executed in the individual operating systems may not be consumed completely. While one operating system is idle, the other operating system may be slacking due to the lack of resource. Such a situation is described with reference to FIG. 4 .

FIG. 4 is a diagram illustrating behaviors of operating systems in a method according to related art.

Referring to FIG. 4 , the time resource consumptions of the first operating system (General Purpose Operating System; GPOS) and the second operating system (Real Time Operating System; RTOS) vary according to the executed tasks. However, the time resources allocated to the respective operating systems are fixed to a predetermined amount. Accordingly, although the first operating system consumes a large amount of time while the second operating system consumes a small amount of time, there is no way to reallocate the remaining time resource allocated to the second operating system without consumption to the first operating system. Also, when the first operating system consumes a small amount of time while the second operating system uses a large amount of time, it is impossible for the second operating system to use the time resource allocated to the first operating system but not consumed.

Aspects of the present invention are to address at least the above-mentioned problems and/or disadvantages and to provide at least the advantages described below. Accordingly, an aspect of the present invention is to provide a method and system for allocating time resource to the respective operating systems dynamically.

In accordance with an aspect of the present invention, a method for allocating resources for a virtualization system supporting at least two operating systems is provided. The method includes allocating resources to the operating systems, calculating, when one of the operating systems is running, workloads for each operating system, and adjusting resources allocated to the operating systems according to the calculated workloads.

In accordance with another aspect of the present invention, a system for allocating resources is provided. The system includes a scheduler for allocating resources to at least two operating systems, and a weight controller for calculating, when the operating systems are running, workloads for each operating system and for adjusting resources allocated to the operating systems according to the calculated workloads.

In accordance with another aspect of the present invention, a computer is provided. The computer includes a multi core processor including a plurality of cores, at least two operating systems, and virtualization software for controlling the simultaneous operation of the at least two operating systems on the multi core processor. The virtualization software includes a scheduler for allocating resources to the at least two operating systems, and a weight controller for calculating workloads for each operating system when the operating systems are running, and for adjusting the resources allocated to the operating systems according to the calculated workloads.

Other aspects, advantages, and salient features of the invention will become apparent to those skilled in the art from the following detailed description, which, taken in conjunction with the annexed drawings, discloses exemplary embodiments of the invention.

The above and other aspects, features, and advantages of certain exemplary embodiments of the present invention will be more apparent from the following description taken in conjunction with the accompanying drawings, in which:

FIG. 1 is a diagram illustrating a configuration of a virtualization system according to related art;

FIG. 2 is a diagram illustrating the principle of a method for a plurality of operating systems to share a resource according to related art;

FIG. 3 is a flowchart illustrating a method for an operating system to operate with allocated time resources according to related art;

FIG. 4 is a diagram illustrating behaviors of operating systems in a method according to related art;

FIG. 5 is a diagram illustrating a principle of a resource allocation method according to an exemplary embodiment of the present invention;

FIG. 6 is a block diagram illustrating a configuration of a virtualization system according to an exemplary embodiment of the present invention;

FIG. 7 is diagram illustrating a method for measuring a workload of operating systems according to a first exemplary embodiment of the present invention;

FIG. 8 is a diagram illustrating a principle of determining a workload of an operating system according to the first exemplary embodiment of the present invention;

FIG. 9 is a flowchart illustrating a method for measuring a workload of the operating system according to a second exemplary embodiment of the present invention;

FIG. 10 is a diagram illustrating a method for adjusting weights of time resources of operating systems according to an exemplary embodiment of the present invention;

FIG. 11 is a flowchart illustrating a method for allocating time resources to operating systems according to an exemplary embodiment of the present invention;

FIG. 12 is a flowchart illustrating a method for adjusting the amounts of time resources allocated to the operating systems depending on workloads according to an exemplary embodiment of the present invention; and

FIG. 13 is a diagram illustrating a principle of adjusting a time resource amount according to an exemplary embodiment of the present invention.

Throughout the drawings, it should be noted that like reference numbers are used to depict the same or similar elements, features, and structures.

The following description with reference to the accompanying drawings is provided to assist in a comprehensive understanding of exemplary embodiments of the invention as defined by the claims and their equivalents. It includes various specific details to assist in that understanding, but these are to be regarded as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the invention. In addition, descriptions of well-known functions and constructions may be omitted for clarity and conciseness.

The terms and words used in the following description and claims are not limited to the bibliographical meanings, but are merely used by the inventor to enable a clear and consistent understanding of the invention. Accordingly, it should be apparent to those skilled in the art that the following description of exemplary embodiments of the present invention is provided for illustration purposes only and not for the purpose of limiting the invention as defined by the appended claims and their equivalents.

It is to be understood that the singular forms “a,” “an,” and “the” include plural referents unless the context clearly dictates otherwise. Thus, for example, reference to “a component surface” includes reference to one or more of such surfaces.

FIG. 5 is a diagram illustrating a principle of a resource allocation method according to an exemplary embodiment of the present invention.

Referring to FIG. 5 , the first operating system (OS 1 ) 510 is allocated the time resource tOS 1 , and the second operating system (OS 2 ) 520 is allocated the time resource tOS 2 . The workload 530 per operating system is measured, and the time resources are allocated dynamically according to the measurement result. FIG. 5 is depicted under the assumption that the workload of OS 1 is small such that the time resource t'OS 1 determined according to the measured workload is smaller than the initially allocated time resource tOS 1 . Although not illustrated, if OS 1 operates under high workload, the time resource t'OS 1 determined according to the measured workload may be larger than the initially allocated time resource tOS 1 .

A virtualization system for allocating resources for each operating system depending on the measured workload is described below with reference to FIG. 6 .

FIG. 6 is a block diagram illustrating a configuration of a virtualization system according to an exemplary embodiment of the present invention.

Referring to FIG. 6 , the virtualization system supporting at least two operating systems includes hardware 610 , virtualization software 620 , and operating systems 630 .

The hardware 610 includes a plurality of cores 613 , a memory 615 , and peripheral devices 617 . The cores 613 are the main processing circuits responsible for processing functions of the virtualization system. The cores 613 include a Performance Monitoring Unit (PMU) 619 for measuring the workload of the operating systems 630 . The PMU 619 calculates the power consumed by the system. The PMU 619 may count the number signal transmissions from the cores 613 to the components of the hardware 610 and calculate the power consumption in accordance with the memory usage.

The memory 615 stores the data generated as a result of function execution in the virtualization system and the programs for executing functions. The peripheral devices 617 are devices connected to the virtualization system.

The virtualization software 620 is configured such that at least two operating systems 630 are running on the hardware 610 simultaneously. The virtualization software 620 includes a plurality of virtual cores 623 , a scheduler 625 , and a weight controller 627 .

A virtual core 623 denotes a core running on the virtualization layer. The operating systems 630 identify the virtual cores 623 as the cores of the hardware 610 in the virtualization system. The scheduler 625 allocates resources for each operating system and schedules the virtual cores 623 for the operation of the operating systems 630 . The scheduler 625 may allocate the time resources to the virtual cores 623 differently according to the workload of the operating system 630 running on the virtual core 623 .

The workload of the operating system 630 is calculated by the weight controller 627 . The weight controller 627 may measure the workload by monitoring the running state of the operating system 630 . The weight controller 627 may measure the workload based on whether the operating system 630 is executing a task. A method for the weight controller 627 to measure the workload is described below with reference to FIGS. 7 to 9 .

The operating system 630 manages the system hardware 610 and provides the hardware virtualization platform and common system services for executing application software. The operating systems 630 run on the virtual core 623 included in the virtualization software 620 .

In the above structured virtualization system in which at least two operating systems are running, the workload is measured for each operating system, and the weight for the time resources to be allocated to each operating system is determined according to the measured workload. The description is made under the assumption that the first and second operating systems are running on the virtualization system. If the workload of the first operating system is greater than that of the second operating system, the virtualization system adjusts the weight of the time resources to be allocated to the second operating system so as to increase the weight of the time resources allocated to the first operating system. If the workload of the first operating system is less than that of the second operating system, the virtualization system adjusts the weight of the time resources allocated to the first operating system so as to increase the weight of the time resources allocated to the second operating system. A method for measuring the workload of each operating system is described below with reference to FIGS. 7 to 9 .

The workload may be calculated in two ways according to the running states of the operating systems. The first method is to notify the weight controller of the idle state periods of the operating systems. The second method is to use the PMU of each core. The method for measuring the workload based on the idle state period of the operating system is described below with reference to FIGS. 7 and 8 .

FIG. 7 is diagram illustrating a method for measuring a workload of operating systems according to a first exemplary embodiment of the present invention.

Referring to FIG. 7 , each operating system notifies the weight controller 627 of the time of the start or end of the idle state. While the operating systems are executing tasks according to the scheduling of the scheduler 625 , the scheduler sends the weight controller 627 a message indicating the end of an idle state and the starting of a task in the corresponding operating system according to the scheduling order. If the operating system enters the idle state according to the scheduling of the scheduler 615 , the scheduler sends the weight controller 627 a message indicating that the operating system has entered the idle state. The weight controller 627 determines the workload of the operating system as depicted in FIG. 8 .

FIG. 8 is a diagram illustrating a principle of determining a workload of an operating system according to the first exemplary embodiment of the present invention.

Referring to FIG. 8 , the weight controller 627 determines the start and end times of the idle state of each operating system according to the message from the scheduler. The weight controller 627 determines the idle state period of the operating system (tidle) 810 based on the start and end time points of the idle state.

The weight controller 627 calculates the workload of the operating system using the idle state period (tidle) 810 and the time resource (tos) allocated to the operating system. The weight controller 627 calculates the workload of the operating system using Equation (1):

workload=( tos−tidle )/ tos   Equation (1)

Using the workload calculated by Equation (1), the weight controller 627 can adjust the time resources to be allocated to each operating system by controlling the scheduler 625 .

A method for measuring a workload using the PMU of each core is described below with reference to FIG. 9 .

FIG. 9 is a flowchart illustrating a method for measuring a workload of an operating system according to a second exemplary embodiment of the present invention.

Referring to FIG. 9 , the weight controller 627 saves the PMU values estimated when the operating system starts running according to the scheduling of the scheduler 625 in step 910 . The PMU value includes the number of times of signal transmissions from the core 613 to the memory 615 and the memory usage amount.

The weight controller determines the running state of the operating systems according to the scheduling of the scheduler 625 in step 920 . The weight controller determines whether the corresponding operating system is executing a task according to the scheduling. If the operating system has completed the execution of the task, the weight controller 627 updates the PMU values by reflecting the completion of the task execution in the operating system in step 930 . The weight controller calculates the workload of the operating system using the initially stored PMU values and the updated PMU values in step 940 . The workload calculation may be performed using a combination of some values of the PMU values, and the PMU values can be changed according to the operating status of each system. The description is made under the assumption that the workload is calculated with the number of memory transactions and the memory usage as PMU values.

Once the number of signal transmissions to the memory and the memory usage are estimated as PMU values, the weight controller 627 calculates the workload of the operating system using Equation (2):

ω×( t )=α· C ( t )+β· M ( t )+γ N ( t )  Equation (2)

where ω×(t) denotes workload, C(t) denotes a system clock cycle, M(t) denotes a number of memory transactions, and N(t) denotes memory usage. The parameters α, β, and γ are threshold values that are selectively used for the system.

After calculating the workloads of the respective operating systems as described above, the weight controller 627 adjusts the weights of the time resources allocated to the operating systems by controlling the scheduler 625 . A method for adjusting the weight of the time resources is described below with reference to FIG. 10 .

FIG. 10 is a diagram illustrating a method for adjusting the weights of the time resources of the operating systems according to an exemplary embodiment of the present invention.

Referring to FIG. 10 , Tos 1 _min denotes a minimum time resource available to be allocated to the first operating system, and Tos 2 _min denotes a minimum time resource available to be allocated to the second operating system. Tos 1 _max denotes a maximum time resource available to be allocated to the first operating system, and Tos 2 _max denotes a maximum time resource available to be allocated to the second operating system. Tos_m denotes the middle value of the weight of the total time resource available for the first and second operating systems.

Tos 1 denotes the time resources used by the first operating system executing a task, and Tos 2 denotes the time resources used by the second operating system executing a task. The time resources to be used by the operating systems may be adjusted according to the workloads of the operating systems. Tos 1 may be adjusted by Tos 1 −Δt or Tos 1 +Δt according to the workload of the first operating system. Tos 2 can be adjusted by Tos 2 −Δt or Tos 2 +Δt according to the workloads of the operating systems. Δt is the adjustment threshold value for minimally adjusting the time resources according to the workload of each operating system. The adjustment threshold value may be set at the virtualization system manufacturing stage and according to the operating environment of the virtualization system. −Δt denotes a decrease of time resources by Δt as compared to the initially allocated time resources of the corresponding operating system, and +Δt denotes an increase of time resources by Δt as compared to the initially allocated time resources of the corresponding operating system.

A method for adjusting the weight of time resources based on the workload of the operating system is described below with reference to FIG. 11 .

FIG. 11 is a flowchart illustrating a method for allocating time resources to operating systems according to an exemplary embodiment of the present invention.

Referring to FIG. 11 , the virtualization system executes the VMM for running at least two operating systems in step 1110 . The virtualization system creates the first operating system OS 1 in step 1120 and the second operating system OS 2 in step 1130 . The virtualization system adjusts the amount of the time resources allocated to the respective OS 1 and OS 2 in step 1140 .

After the time resource allocation, the virtualization system determines whether the scheduler is scheduling OS 1 operation or OS 2 operation in step 1150 . If the OS 1 operation is scheduled, the virtualization system executes the operation of OS 1 in step 1160 . If the OS 2 operation is scheduled, the virtualization system executes the operation of OS 2 in step 1170 . After completing the OS operation, the virtualization system controls the weight controller to adjust the time resources allocated to the respective OSs in step 1180 . Step 1180 of FIG. 11 is described below with reference to FIG. 12 . After adjusting the time resources allocated to the respective OSs, the virtualization system repeats the procedure from step 1150 .

FIG. 12 is a flowchart illustrating a method for adjusting amounts of time resources allocated to operating systems depending on workloads according to an exemplary embodiment of the present invention.

Referring to FIG. 12 , the virtualization system calculates the workload (W) of each operating system in step 1205 . It is assumed that the minimum allocation time resources, Tos 1 _min and Tos 2 _min, are allocated as the initial time resources of the respective operating systems. The virtualization system determines whether the workload of OS 1 (Wos 1 ) has decreased as compared to the previous workload in step 1210 . If Wos 1 has decreased, the virtualization system adjusts the weight of the time resources allocated to OS 1 (Tos 1 ) by −Δt in step 1213 . The virtualization system determines whether the adjusted Tos 1 is less than Tos 1 _min in step 1215 . If the adjusted Tos 1 is less than Tos 1 _min, the virtualization system sets Tos 1 equal to Tos 1 _min in step 1217 . The reason for setting Tos 1 equal to Tos 1 _min is to guarantee the interactivity of OS 1 by guaranteeing a minimum operation of OS 1 .

If Wos 1 has not decreased in step 1210 , the virtualization system determines whether Wos 1 has increased as compared to the previous workload in step 1220 . If Wos 1 has increased, the virtualization system determines whether Tos 1 +Δt is equal to or greater than Tos 1 _max in step 1223 .

If Tos 1 +Δt is less than Tos 1 _max, the virtualization system increases Tos 1 by +Δt. If Tos 1 +Δt is equal to or greater than Tos 1 _max, the virtualization system determines whether the valued obtained by subtracting the current consumption of Tos 2 from the maximum time resources allocated to OS 2 (Tos 2 _max) is equal to or greater than Δt in step 1227 . If Tos 2 _max-Tos 2 is equal to or greater than Δt, the virtualization system increases the Tos 1 _max by Δt in step 1229 . The virtualization system sets Tos 1 _max equal to Tos 1 which has increased by +Δt and decreases Tos 2 _max by −Δt in step 1230 .

If Tos 2 _max−Tos 2 is less than Δt in step 1227 , the virtualization system determines whether Tos 1 _max is less than Tos_m in step 1233 . If Tos 1 _max is less than Tos_m, the virtualization increases Tos 1 by +Δt and decreases Tos 2 by −Δt in step 1235 . The virtualization system sets Tos 1 _max equal to Tos 1 +Δt and decreases Tos 2 by −Δt in step 1237 . Reference numbers {circle around ( 1 )}, {circle around ( 2 )}, {circle around ( 3 )}, {circle around ( 4 )}, {circle around ( 5 )}, and {circle around ( 6 )} of FIG. 12 are provided for cross reference with FIG. 13 .

The weight of the time resources adjusted according to the time resource allocation method of FIG. 12 is described below with reference to FIG. 13 .

FIG. 13 is a diagram illustrating a principle of adjusting a time resource amount according to an exemplary embodiment of the present invention. In FIGS. 12 and 13 , OS 1 and OS 2 are allocated the time resources of Tos 1 _min and Tos 2 _min respectively. The workload of OS 1 changes while OS 1 is running such that Tos 1 is adjusted by Δt.

Referring to FIG. 13 , part {circle around ( 1 )} of FIG. 13 shows an exemplary case where, if Tos 1 is decreased by −Δt due to the reduced workload of OS 1 and if the decreased Tos 1 is less than Tos 1 _min, Tos 1 is set equal to Tos 1 _min. Part {circle around ( 2 )} of FIG. 13 shows an exemplary case where Tos 1 is decreased by −Δt due to the reduced workload of OS 1 . Part {circle around ( 3 )} of FIG. 13 shows an exemplary case where Tos 1 increases by +Δt due to the increased workload of OS 1 . Part {circle around ( 4 )} of FIG. 13 shows an exemplary case where Tos 1 increased by +Δt is greater than Tos 1 _max due to the increased workload of OS 1 . Tos 1 may be increased or maintained according to the variation of workload of OS 2 . If the difference between Tos 2 and Tos 2 _max is greater than Δt when Tos 1 has reached Tos 1 _max, it is possible to decrease Tos 1 _max by −Δt and increase Tos 2 and Tos 2 _max by Δt.

Part {circle around ( 5 )} of FIG. 13 shows an exemplary case where Tos 2 _max is decreased and Tos 1 _max is increased due to the increased workload for Tos 1 in a state that Tos 1 _max has moved to Tos 1 . Part {circle around ( 6 )} of FIG. 13 shows an exemplary case where Tos 1 _max and Tos 1 are maintained in a state where it is impossible to decrease Tos 2 _max.

Although the description has been directed to a method for adjusting the time resources allocated to OS 1 in adaptation to the workload of OS 1 , the method may be applied for the other OS. The time resources allocated to OS 2 may be adjusted according to the workload of OS 2 .

As described above, the resource allocation method according to exemplary embodiments of the present invention is capable of allocating time resources dynamically to multiple operating systems running in the virtualization system. The resource allocation method of the present invention allocates the time resources to the multiple operating systems dynamically so as to minimize resource waste and inefficiency. As a consequence, the resource allocation method according to exemplary embodiments of the present invention is capable of facilitating operations of multiple operating systems running in the virtualization system and reducing power consumption of the virtualization system.

While the invention has been shown and described with reference to certain exemplary embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims and their equivalents.