TW201903608A - Virtual network system and virtual network system communication method - Google Patents

Abstract

The purpose of the present invention is to realize easy communications between virtual machines, without relying on a proprietary application running on a virtual machine or a virtual communication device on a hypervisor. To that end, in the present invention a virtual communication device (4b) included in a virtual machine guest (1b) of a transmission source writes communication data if a shared memory (61b) is available, and issues, by a hypervisor call, a notification of the completion of the communication data write. A hypervisor (9) issues an interrupt (X) to a virtual machine guest (1a) of an addressee. A virtual communication device (4a), triggered by the interrupt (X), reads the communication data written to the shared memory (61b), and issues, by a hypervisor call, a notification of the completion of the communication data read. The hypervisor (9) issues an interrupt (Y) to the virtual machine guest (1b) of the transmission source. Triggered by the interrupt (Y), the virtual communication device (4b) detects availability in the shared memory (61b).

Description

 Virtual network system and communication method of virtual network system  

The present invention relates to a communication method for connecting a virtual machine embodied on a hypervisor into a virtual network system capable of communication and a virtual network system.

In recent years, the processing performance of physical computers has been increasing. It is generally followed by the hypervisor acting on the physical computer to materialize the plurality of virtual machines. The information exchange between the virtual machines is realized by the control of the virtual communication device on the hypervisor by the virtualized input/output interface. Such a virtualized input and output interface is complicated and a large overhead is incurred.

It can be considered that by embodying the virtual communication device on the virtual machine, the control from the guest operating system (Guest Operating System) operating on the virtual machine can be simplified, and the additional burden can be eliminated. As an example, there is an invention described in Patent Document 1.

In the solution of the abstract of Patent Document 1, it has been described that "the guest OS OS request packet is received by the virtual network interface card (NIC) driver 308 of the guest OS 210 executed by the guest virtual machine 206. (request packet), and a block is added to a transmission queue 506 in a shared memory 504 of a computing system, thereby sharing a physical NIC in a plurality of virtual machines. Then, the server virtual machine 214 fetches the block from the sending queue by the virtual NIC driver 316 of the server OS 216, and packages the server OS request packet and delivers it to the server. The server OS network stack portion 314 of the server OS is routed to the entity NIC driver 318 by a bridge driver 502 and sends a server OS request packet on the network. "."

[Previous Technical Literature]

[Patent Literature]

Patent Document 1: Japanese Laid-Open Patent Publication No. 2012-003747.

According to the invention described in Patent Document 1, a queue is created on a shared memory, and communication between virtual machines is realized by the queue. However, in order to achieve such communication, it is necessary to prepare a separate application, and more complicated processing for realizing the queue.

Accordingly, the subject of the present invention is to enable virtual network systems to easily communicate between virtual machines without relying on a single application operating on a virtual machine or a virtual communication device on a hypervisor.

The present invention provides a virtual network system, including: a hypervisor; the first virtual machine is embodied by the foregoing hypervisor and includes a virtual communication device; the second virtual machine is borrowed Specifically embodied by the hypervisor, including a virtual communication device; and shared memory shared by the first virtual machine and the second virtual machine.

The virtual communication device of the virtual device included in the transmission source in the first virtual machine and the second virtual device is written in the communication material as long as the shared memory is empty, and is included in the communication by the hypervisor call The virtual communication device of the virtual machine of the destination notifies the completion of the writing of the communication data.

The aforementioned hypervisor causes the virtual machine of the aforementioned destination to generate a first interrupt.

The virtual communication device of the virtual machine included in the destination reads the communication data written to the shared memory by using the first interrupt, and is included in the transmission source by the hypervisor call. The virtual communication device of the virtual machine notifies the completion of the reading of the communication data.

The aforementioned hypervisor causes the virtual machine of the aforementioned transmission source to generate a second interrupt.

The virtual communication device of the virtual machine included in the transmission source detects that the shared memory has become empty by using the second interrupt.

According to the present invention, for a virtual network system, communication between virtual machines can be easily realized without relying on a single application operating on a virtual machine or a virtual communication device on a hypervisor.

1, 1a to 1n‧‧‧ virtual machine object (an example of a virtual machine)

2‧‧‧Switch

3, 3a, 3b‧‧‧ virtual communication device driver

4, 4a, 4b‧‧‧ virtual communication device

5, 5b‧‧‧ web application

8‧‧‧External network

9‧‧‧Super Manager

11, 11a to 11n‧‧‧ Guest OS

41, 41a, 41b‧‧‧ virtual register

61, 61b to 61n,

62, 62b to 62n‧‧‧ shared memory

71‧‧‧Communication device driver

72‧‧‧Communication device

S‧‧‧Virtual Network System

Fig. 1 is a view showing the configuration of a virtual network system in the first embodiment.

Figure 2 is a communication sequence showing the communication frame to the read memory of the shared memory.

Figure 3 is a communication sequence informing the virtual machine that the readout is complete.

Fig. 4 is a view showing a communication sequence in which a communication frame is written to read from a shared memory in a variation of the first embodiment.

Figure 5 is a communication sequence informing the virtual machine that the readout is complete.

Fig. 6 is a view showing the configuration of a virtual network system in the second embodiment.

Figure 7 shows the communication sequence of the communication frame to the shared memory.

Figure 8 is a communication sequence informing the virtual machine that the readout is complete.

Hereinafter, embodiments for carrying out the invention will be described in detail with reference to the drawings.

Fig. 1 is a view showing the configuration of a virtual network system S in the first embodiment.

As shown in FIG. 1, the virtual network system S includes a hypervisor 9, virtual machine objects 1a to 1n, shared memories 61b to 61n, and shared memories 62b to 62n. The hypervisor 9 and the virtual machine objects 1a to 1n are embodied by a CPU (Central Processing Unit) of a server device (not shown) to execute a software program. The virtual machine objects 1a to 1n refer to virtual machines embodied by the hypervisor 9. The virtual machine objects 1a to 1n further operate the guest OSs 11a to 11n.

The server device (not shown) that embody the virtual network system S includes a physical communication device 72 in addition to the CPU and the memory. A widely known system firmware as a basic input/output system (BIOS: Basic Input Output System) is executed by a server device. The BIOS starts the specification, testing, and initialization of the components of the server device at the system boot time, and provides an interface between the hardware and software of the server device. According to an embodiment of the present invention, after the BIOS has just completed initialization of the server device, the hypervisor program is executed and the hypervisor 9 of FIG. 1 is embodied.

The hypervisor 9 simultaneously executes a plurality of virtual machine objects 1a to 1n on the server device. Further, the hypervisor 9 monitors the execution of the guest OSs 11a to 11n as the guest operating system.

Hereinafter, when each of the virtual machine objects 1a to 1n is not particularly distinguished, it is simply described as the virtual machine object 1. When the individual guest OSs 11a to 11n are not particularly distinguished, they are simply described as the guest OS 11. Further, in the drawings and the like, there are cases where the virtual machine object 1a is described as "virtual machine #1" or "VM#1". Furthermore, there are cases where the virtual machine object 1b is described as "virtual machine #2" or "VM#2". Similarly, in some cases, the virtual machine object 1c is described as "virtual machine #3" or "VM#3", and the virtual machine object 1n is described as "virtual machine #n" or "VM#n".

The virtual machine objects 1a to 1n share a hardware resource of the server device. Although even multiple applications share hardware resources in multitasking, they cannot separate an application error from other applications. On the other hand, in the virtual machine objects 1a to 1b using the hypervisor 9, an error in one operating system can be separated from another operating system sharing the hardware. The hypervisor 9 is an execution environment for preparing virtual machine objects 1a to 1n in the system. The hypervisor 9 activates more than one virtual machine object 1 as needed. The virtual machine object 1 system further executes the guest OS 11.

The hypervisor 9 is capable of simultaneously starting and executing a plurality of virtual machine objects 1a to 1n. The virtual machine object 1a (first virtual machine) includes a guest OS 11a and a virtual communication device 4a. The virtual communication device 4a is a virtual communication device constructed on the virtual machine object 1a, and is controlled by a virtual communication device driver 3a as will be described later.

The guest OS 11a refers to an operating system that operates on the virtual machine object 1a, and is a system software for operating or using the virtual machine object 1a. Furthermore, the guest OS 11a is a virtual communication device driver 3a, a switch 2, and a physical communication device driver 71. The virtual communication device driver 3a is a program for controlling the virtual communication device 4a. The switch 2 refers to the switch of the relay communication frame, and refers to the MAC (Media Access Control) address to determine the destination of the communication frame and relays the L2 (Layer 2) ; second layer) exchanger. The virtual communication device 4a has different MAC addresses.

The virtual machine object 1a relays communication between the virtual machine objects 1b to 1n and the external network 8 by the switch 2.

The virtual machine object 1b (an example of the second virtual machine) includes a guest OS 11b and a virtual communication device 4b. The virtual communication device 4b is a virtual communication device constructed on the virtual machine object 1b, and is controlled by the virtual communication device driver 3b described later. The guest OS 11b refers to an operating system that operates on the virtual machine object 1b, and is a system software for operating or using the virtual machine object 1b.

Furthermore, the guest OS 11b is a virtual communication device driver 3b and a network application 5b. The virtual communication device driver 3b is a program for controlling the virtual communication device 4b. The web application 5b refers to an application software used by the guest OS 11b to facilitate users. The virtual machine objects 1c to 1n (an example of the second virtual machine) are also configured in the same manner as the virtual machine object 1b.

Hereinafter, when the virtual communication devices 4a, 4b, ... included in the respective virtual machine objects 1a to 1n are not particularly distinguished, they are simply described as the virtual communication device 4. When the virtual communication device drivers 3a, 3b, ... included in the respective guest OSs 11a to 11n are not particularly distinguished, they are simply described as the virtual communication device driver 3. When the web applications 5b, ... included in the respective guest OSs 11b to 11n are not particularly distinguished, they are simply described as the web application 5.

The virtual communication device 4 included in the virtual machine objects 1b to 1n transmits a communication frame to the virtual communication device 4a included in the virtual machine object 1a via the shared memory 61b to 61n. Further, the virtual communication device 4 included in the virtual machine objects 1b to 1n receives the communication frame from the virtual communication device 4a via the shared memory 62b to 62n. The shared memories 61b to 61n are memory areas that can be written from the virtual communication devices 4 included in the respective virtual machine objects 1b to 1n, and are readable from the virtual communication device 4a included in the virtual machine object 1a. Memory area. The shared memory 62b to 62n is a memory area that can be written from the virtual communication device 4a included in the virtual machine object 1a, and is a memory that can be read from the virtual communication device 4 included in each of the virtual machine objects 1b to 1n. Body area.

In the following, when the shared memory bodies 61b to 61n are not particularly distinguished, the shared memory 61 is simply described, and when the shared memory bodies 62b to 62n are not particularly distinguished, the shared memory 62 is simply described.

Furthermore, the virtual communication device 4 can also be provided with a queue buffer for transmission and reception. At this time, the virtual communication device 4 of the transmission source does not wait for the completion of the reading of the communication frame by the destination virtual communication device 4 from the virtual communication device as long as there is a space in the transmission buffer. Driver 3 receives the communication frame. Similarly, the virtual communication device 4 on the receiving side does not wait for the virtual input from the transmission source to wait for reception of the communication frame by the virtual communication device driver 3 as long as there is a space in the buffer for reception. The communication frame of the communication device 4.

According to the first embodiment, in order to provide network services to the respective web applications 5, the hypervisor 9 embodyes the virtual communication device 4 for each virtual machine object of the virtual machine objects 1a to 1n. For the virtual machine objects 1a to 1n, the virtual communication device 4 has the same function as the physical communication device 72. Each virtual communication device 4 processes the network requirements from the higher-order virtual communication device driver 3 and has the task of returning the result of the request. For example, the virtual communication device 4b included in the virtual machine object 1b transmits a communication frame to the virtual communication device 4a via the shared memory 61b. The virtual communication device 4b included in the virtual machine object 1b receives the communication frame from the virtual communication device 4a through the shared memory 62b.

Similarly, the virtual communication device 4 included in the virtual machine objects 1c to 1n transmits a communication frame to the virtual communication device 4a via the shared memory 61c to 61n. The virtual communication device 4 included in the virtual machine objects 1c to 1n further receives the communication frame from the virtual communication device 4a through the shared memory 62c to 62n.

The virtual network of the first embodiment constitutes a topology of a star connection with the virtual machine object 1a as a center. The virtual machine object 1a relays the communication frame with the external network 8 by the switch 2 connected to the virtual communication device driver 3a and the physical communication device driver 71.

The virtual machine object 1a exchanges information between the virtual machine objects 1b to 1n by virtual machine-to-machine communication.

In the virtual machine-to-device communication of the first embodiment, the virtual machine object 1 of the transmission source writes a communication frame to the shared memory 61 or the shared memory 62. Furthermore, the virtual machine object 1 notifies the hypervisor 9 of the completion of the write by the hypervisor call. The hypervisor 9 notifies the virtual machine object 1 of the destination of the completion of the writing of the communication frame by interrupt X. The virtual machine object 1 of the destination receives the notification by the interrupt X and reads out the communication frame on the shared memory 61 or the shared memory 62.

Further, the virtual machine object 1 of the destination notifies the hypervisor 9 that the reading is completed by the hypervisor call. The hypervisor 9 notifies the virtual machine object 1 of the transmission source that the reading of the communication frame is completed by interrupting Y. The virtual machine-to-machine communication realizes two-way communication by symmetrically having the above configuration.

Furthermore, it is a star connection in which the virtual machine object 1a exchanges information with the virtual machine objects 1b to 1n. Although the virtual machine objects 1b to 1n are sufficient as long as there are interrupt X and interrupt Y, the virtual machine object 1a needs to interrupt X _B to X _n and interrupt Y _B to Y _n .

The sequence in which the virtual communication device 4 uses these structures and transmits the communication frame from the virtual machine object 1b to the virtual machine object 1a is shown in FIGS. 2 and 3 below. Furthermore, the order of transmitting the communication frame from the virtual machine object 1a to the virtual machine object 1b is also the same.

Figures 2 and 3 are communication sequences between virtual machines. The communication sequence of FIG. 2 shows the write-to-read of the communication frame to the shared memory 61b.

In step S10, the virtual communication device 4b included in the virtual machine object 1b waits in the original state as long as the shared memory 61b is not empty ("NO"), and if the shared memory 61b is empty ("Yes"), The process proceeds to step S11. Furthermore, the shared memory 61b is empty in the initial state.

In step S11, when the virtual communication device 4b receives the communication frame from the virtual communication device driver 3b of the higher layer, it writes it to the shared memory 61b (step S12). The shared memory 61b is written into the communication frame and is not empty. Moreover, the virtual communication device 4b notifies the writing of the completed communication frame and the length information of the communication frame by the hypervisor call (step S13). Furthermore, in the drawing, the hypervisor call is described as "HVC" (Hypervisor call). The communication frame is, for example, sent by the web application 5b to the external network 8. Furthermore, the length information of the communication frame is not necessary, but can also be set to a fixed length.

Next, the hypervisor 9 recognizes that the virtual machine object 1b has completed writing to the shared memory 61b from the hypervisor call in step S13, and acquires the length information of the communication frame (step S14). The hypervisor 9 causes the virtual machine object 1a to generate an interrupt X (step S15). The virtual communication device 4a included in the virtual machine object 1a recognizes that the virtual machine object 1b has completed writing to the shared memory 61b by the interrupt X (step S16).

The virtual communication device 4a included in the virtual machine object 1a requests the length information of the communication frame written to the shared memory 61b by the hypervisor call (step S17). The hypervisor 9 returns the length information of the communication frame (step S18). The virtual communication device 4a reads out the communication frame that has been written to the shared memory 61b (step S19), and delivers the communication frame to the virtual communication device driver 3a of the upper layer (step S20). Thereafter, the communication frame is delivered to the switch 2 through the virtual communication device driver 3a.

Figure 3 is a communication sequence informing the virtual machine that the readout is complete.

As shown in step S21 of Fig. 3, the virtual communication device 4a notifies that the reading of the shared memory 61b has been completed by the hypervisor call. When the hypervisor 9 recognizes that the virtual machine object 1a has completed reading by the hypervisor call of step S21 (step S22), the virtual machine object 1b is caused to generate an interrupt Y (step S23). Thereby, the virtual communication device 4b recognizes that the reading by the virtual machine object 1a is completed (step S24), and recognizes that the shared memory 61b is empty (step S25). Thereafter, the processing from step S10 onward can be performed again.

Next, the inter-virtual machine communication in the variation of the first embodiment will be described. The virtual machine object 1 of the transmission source writes a communication frame to the shared memory 61 or the shared memory 62, and notifies the completion of the write by the hypervisor call. The hypervisor 9 notifies the virtual machine object 1 of the destination of a communication event by interrupt X. The virtual machine object 1 of the destination, when receiving the notification by the interrupt X, further obtains the interrupt factor and the frame length information by the hypervisor call. Since the interrupt is caused by the completion of the writing of the communication frame, the virtual machine object 1 of the destination reads the communication frame located in the shared memory 61 or the shared memory 62.

Moreover, the virtual machine object 1 of the destination notifies the hypervisor 9 that the reading is completed by the hypervisor call. The hypervisor 9 notifies the virtual machine object 1 of the transmission source of the occurrence of the communication event by the interrupt X. The virtual machine object 1 of the transmission source receives the notification by the interrupt X, and further obtains the interruption factor by the hypervisor call. Since the interrupt is caused by the completion of the reading of the communication frame, the virtual machine object 1 of the transmission source recognizes that the reading of the communication frame is completed. The virtual machine-to-machine communication realizes two-way communication by symmetrically having this configuration.

The sequence in which the virtual communication device 4 uses these structures and transmits the communication frame from the virtual machine object 1b to the virtual machine object 1a is shown in Figs. 4 and 5 below. Furthermore, the order of transmitting the communication frame from the virtual machine object 1a to the virtual machine object 1b is also the same.

4 and 5 show a communication sequence between virtual machines in a variation of the first embodiment. The communication sequence of FIG. 4 shows the write-to-read of the shared frame 61b by the communication frame.

In step S30, the virtual communication device 4b included in the virtual machine object 1b waits in the original state as long as the shared memory 61b is not empty ("NO"), and if the shared memory 61b is empty ("Yes"), The process proceeds to step S31.

In step S31, when the virtual communication device 4b receives the communication frame from the virtual communication device driver 3b of the higher layer, it writes it to the shared memory 61b (step S32). A communication frame is written in the shared memory 61b, and is not empty. Moreover, the virtual communication device 4b notifies the writing of the completed communication frame and the length information of the communication frame by calling the hypervisor (step S33).

Next, the hypervisor 9 recognizes that the virtual machine object 1b has completed writing to the shared memory 61b from the hypervisor call of step S33, and acquires the length information of the communication frame (step S34). The hypervisor 9 sets the write completion information (step S35), and causes the virtual machine object 1a to generate the interrupt X (step S36). The virtual communication device 4a included in the virtual machine object 1a inquires of the interrupt cause by the hypervisor call (step S37). In response to this, the hypervisor 9 responds to the writing completion and communication frame length information for the shared memory 61b by the virtual machine object 1a (step S38). Thereby, the virtual communication device 4a recognizes that the virtual machine object 1b has completed writing to the shared memory 61b (step S39).

The virtual communication device 4a re-sets the write completion information of the virtual machine object 1a by the hypervisor call (step S40). The virtual communication device 4a reads out the communication frame that has been written to the shared memory 61b (step S40), and delivers the communication frame to the virtual communication device driver 3a of the upper layer (step S42). Thereafter, the communication frame is delivered to the switch 2 through the virtual communication device driver 3a.

Figure 5 is a communication sequence informing the virtual machine that the readout is complete.

As shown in step S43 of Fig. 5, the virtual communication device 4a notifies that the reading of the shared memory 61b has been completed by the hypervisor call. When the hypervisor 9 recognizes that the reading of the virtual machine object 1a is completed by the hypervisor call of step S43 (step S44), after the read completion information is set (step S45), the virtual machine object 1b is made. An interrupt X is generated (step S46). Thereby, the virtual communication device 4b included in the virtual machine object 1b inquires of the interrupt cause by the hypervisor call (step S47). In contrast, the hypervisor 9 responds to the completion of reading of the virtual machine object 1a (step S48). The virtual communication device 4a recognizes that the virtual machine object 1a has completed reading of the shared memory 61b (step S49). Further, the virtual communication device 4a resets the read completion information by the hypervisor call (step S50). Thereafter, the virtual communication device 4a recognizes that the shared memory 61b is empty (step S51). Thereafter, the processing from step S30 onward can be performed again.

In this variation, only one interrupt X is used to notify the completion of writing and the completion of reading in the virtual communication. Thus, the interrupt resources used for virtual communication can be minimized.

This variation is a star connection in which the virtual machine object 1a exchanges information with the virtual machine objects 1b to 1n. The virtual machine objects 1b to 1n are sufficient as long as there is an interrupt X, but the virtual machine object 1a requires the same number of interrupts X _B to X _n as the number of communication docking ends.

Fig. 6 is a view showing the configuration of the virtual network system S in the second embodiment.

The virtual communication device 4a of the virtual network system S of the second embodiment includes a virtual register 41a, and the virtual communication device 4b includes a virtual register 41b. Hereinafter, when the virtual registers 41a, 41b, ... are not particularly distinguished, they are simply described as the virtual register 41. The virtual register 41 is a register that is written by the hypervisor 9 and read by each virtual communication device 4. The hypervisor 9 can transmit information to each virtual communication device 4 through the virtual register 41.

The virtual network system S of the second embodiment further limits the interruption from the hypervisor 9 to each virtual communication device 4 to only one. This allows you to minimize interrupt resources.

Next, the virtual machine communication of the second embodiment will be described. The virtual machine object 1 of the transmission source writes a communication frame to the shared memory 61 or the shared memory 62, and notifies the super manager 9 of the completion of writing by the hypervisor call. The hypervisor 9 notifies the virtual machine object 1 of the destination of the occurrence of the communication event by interrupting after the communication frame length information and the writing completion information have been set to the virtual register 41.

The virtual machine object 1 of the destination obtains the interrupt factor and the frame length information by the virtual register 41 when receiving the notification by the interrupt. Since the interrupt is caused by the completion of the writing of the communication frame, the virtual machine object 1 of the destination reads the communication frame located in the shared memory 61 or the shared memory 62.

Moreover, the virtual machine object 1 of the destination notifies the hypervisor 9 that the reading is completed by the hypervisor call. The hypervisor 9 notifies the virtual machine object 1 of the transmission source of the occurrence of the communication event by the interruption. The virtual machine object 1 of the transmission source obtains the interrupt factor and the communication frame length information by the virtual register 41 when receiving the notification by the interrupt. Since the interrupt is caused by the completion of the reading of the communication frame, the virtual machine object 1 of the transmission source recognizes that the reading of the communication frame is completed. Virtual machine-to-machine communication achieves two-way communication by symmetrically having this configuration.

Furthermore, it is a star connection in which the virtual machine object 1a exchanges information with the virtual machine objects 1b to 1n. Although it is sufficient that the virtual machine objects 1b to 1n have one interruption, the virtual machine object 1a requires an interruption of the number corresponding to the number of virtual machine objects 1b to 1n of the destination.

The order in which the virtual communication device 4 uses these structures and transmits the communication frame from the virtual machine object 1b to the virtual machine object 1a is shown in FIGS. 7 and 8 below. Furthermore, the order of transmitting the communication frame from the virtual machine object 1a to the virtual machine object 1b is also the same.

7 and 8 are communication sequences between virtual machines in the second embodiment.

In step S60, the virtual communication device 4b included in the virtual machine object 1b waits in the original state as long as the shared memory 61b is not empty ("NO"), and if the shared memory 61b is empty ("Yes"), The process proceeds to step S61. Furthermore, the shared memory 61b is empty in the initial state.

In step S61, when the virtual communication device 4b receives the communication frame from the virtual communication device driver 3b of the higher layer, it writes it to the shared memory 61b (step S62). A communication frame is written in the shared memory 61b, and is not empty. Further, the virtual communication device 4b notifies the length information and the writing completion of the communication frame by the hypervisor call (step S63).

Next, the hypervisor 9 recognizes that the virtual machine object 1b has completed writing to the shared memory 61b from the hypervisor call in step S63, and acquires the length information of the communication frame (step S64). The hypervisor 9 sets the write completion information on the virtual machine object 1b to the virtual register 41a (step S65), and writes the length information of the communication frame to the virtual register 41a (step S66). Thereafter, the hypervisor 9 causes the virtual machine object 1a to be interrupted (step S67). The virtual communication device 4a included in the virtual machine object 1a reads out the virtual register 41, and recognizes that the interrupt factor is completed by the virtual machine object 1b (step S68), and reads the length information of the communication frame. (Step S69). Thereafter, the virtual communication device 4a resets the write completion information of the virtual machine object 1b that has been set in the virtual register 41b (step S70).

The virtual communication device 4a reads out the communication frame that has been written to the shared memory 61b (step S71), and delivers the communication frame to the virtual communication device driver 3a of the upper layer (step S72). Thereafter, the communication frame is delivered to the switch 2 through the virtual communication device driver 3a.

As shown in step S73 of Fig. 8, the virtual communication device 4a notifies the completion of the reading by the hypervisor call. The hypervisor 9 recognizes that the reading of the virtual machine object 1a is completed by the hypervisor call in step S73 (step S74), and the reading of the virtual machine object 1a has been set to the virtual register 41b. After the completion of the information (step S75), the virtual machine object 1b is caused to be interrupted (step S76). Thereby, the virtual communication device 4b included in the virtual machine object 1b recognizes that the interrupt cause is the completion of the reading of the virtual machine object 1a by the virtual register 41b (step S77). Thereafter, the virtual communication device 4b resets the read completion information of the virtual register 41b (step S78). Thereafter, the virtual communication device 4a recognizes that the shared memory 61b is empty (step S79). Thereafter, the processing from step S60 onward can be performed again.

 (variation)  

The present invention is not limited to the above-described embodiments, and can be modified and implemented without departing from the scope of the present invention. For example, the following aspects (a) to (i) are possible.

(a) The sequence of writing and reading of the communication frame and the notification sequence of reading completion shown in each embodiment may be arbitrarily combined. For example, the sequence of the first embodiment shown in FIG. 2 and the notification sequence of the second embodiment shown in FIG. 8 may be combined.

(b) The switch of the above embodiment is not limited to the L2 switch, and may be an L3 (Layer 3; Layer 3) switch or router.

(c) In the above embodiment, the length of the communication frame written to the shared memory 61 or the shared memory 62 may not be a variable length, that is, it may be used as a fixed length.

(d) Each virtual machine may also recognize the completion of writing or completion of reading by means of a polling to the virtual register or a repetition of a hypervisor call without interruption.

(e) In the above embodiment, the information distributed by the hypervisor call or the virtual body register may include metadata not shown in the above embodiment.

(f) In the above embodiment, the virtual communication device or the switch may be placed in the hypervisor. The physical communication device or its driver is also the same. At this point, the hypervisor call or interrupt system can be replaced with a simple function call.

(g) In the first embodiment, the virtual communication device 4a may also set the interruption of communication with the other virtual communication devices 4b, ... to two, and recognize the completed writing by the hypervisor call. The incoming virtual communication device 4 or the virtual communication device 4 that has completed reading.

(h) In the variation of the first embodiment, the virtual communication device 4a may also set the interruption of communication with the other virtual communication devices 4b, ... as one, and identify by the hypervisor call. The cause of the interruption, and the virtual communication device 4 that caused the interruption cause.

(i) In the second embodiment, the virtual communication device 4a may also set an interrupt for communication with the other virtual communication devices 4b, ..., and identify the cause of the interruption by the virtual register. And the virtual communication device 4 that causes the interruption to be generated.

Claims (5)

 A virtual network system includes: a hypervisor; the first virtual machine is embodied by the hypervisor and includes a virtual communication device; and the second virtual machine is embodied by the hypervisor. And including a virtual communication device; and shared memory shared by the first virtual machine and the second virtual machine; and virtual communication of the virtual machine included in the source of the first virtual machine and the second virtual machine The device writes the communication data as long as the shared memory is empty, and notifies the virtual communication device of the virtual machine included in the destination by the hypervisor call to complete the writing of the communication data; the super manager is The virtual machine of the destination destination generates a first interrupt; the virtual communication device of the virtual machine included in the destination reads the communication data written to the shared memory by using the first interrupt as a trigger, and The manager calls to notify the virtual communication device of the virtual machine included in the foregoing transmission source that the communication data is read out; the foregoing super management The virtual machine of the virtual source of the transmission source generates a second interrupt; and the virtual communication device of the virtual machine included in the transmission source detects that the shared memory has become empty by using the second interrupt.    The virtual network system as claimed in claim 1, wherein the virtual communication device of the virtual machine included in the destination obtains the length information of the communication data by using the hypervisor call, and obtains the information by using the first interrupt. The communication data that has been written to the aforementioned shared memory.    The virtual network system as claimed in claim 1, wherein the first virtual machine and the second virtual machine further comprise a virtual register; the super manager is configured to use a virtual machine by using the foregoing sending source. After the hypervisor call recognizes that the writing is completed, the virtual machine of the destination may generate the first interruption after writing the length information of the communication data to the virtual register of the virtual machine of the destination; The virtual communication device of the virtual machine of the destination acquires the length information of the communication data from the virtual register by using the first interrupt, and acquires the communication data written in the shared memory.    The virtual network system as claimed in claim 1, wherein the first virtual machine and the second virtual machine further comprise a virtual register; the super manager is configured to use a virtual machine by the destination. After the hypervisor calls the read completion, the virtual machine of the transmission source generates the first interrupt after the readout of the virtual register setting of the virtual machine of the transmission source is completed; and is included in the foregoing transmission source. The virtual communication device of the virtual machine detects that the shared memory has become empty by the setting of the first interrupt and the readout of the virtual temporary register.    A communication method of a virtual network system is a communication method performed by a virtual network system, the virtual network system includes: a hypervisor; the first virtual machine is embodied by the foregoing hypervisor, and includes a virtual communication device; the second virtual machine is embodied by the hypervisor and includes a virtual communication device; and the shared memory is shared by the first virtual machine and the second virtual machine; the first virtual The virtual communication device of the virtual machine included in the transmission source of the machine and the second virtual machine is written to the communication material as long as the shared memory is empty; and the virtual machine included in the destination is called by the hypervisor call The virtual communication device notifies the completion of the writing of the communication data; the super manager causes the virtual machine of the destination to generate the first interruption; and the virtual communication device of the virtual machine included in the destination is triggered by the first interruption. Reading the communication data that has been written to the aforementioned shared memory; calling the virtual source included in the aforementioned transmission source by the hypervisor call The virtual communication device of the machine notifies that the reading of the communication data is completed; the super manager causes the virtual machine of the transmitting source to generate a second interrupt; and the virtual communication device of the virtual machine included in the transmitting source takes the second interrupt as an opportunity The detection of the aforementioned shared memory has become empty.