CN114281529B - Method, system and terminal for dispatching optimization of distributed virtualized client operating system - Google Patents
Method, system and terminal for dispatching optimization of distributed virtualized client operating system Download PDFInfo
- Publication number
- CN114281529B CN114281529B CN202111508295.XA CN202111508295A CN114281529B CN 114281529 B CN114281529 B CN 114281529B CN 202111508295 A CN202111508295 A CN 202111508295A CN 114281529 B CN114281529 B CN 114281529B
- Authority
- CN
- China
- Prior art keywords
- operating system
- virtual machine
- vcpu
- node
- information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 53
- 238000005457 optimization Methods 0.000 title claims abstract description 25
- 230000003993 interaction Effects 0.000 claims abstract description 54
- 238000004891 communication Methods 0.000 claims description 31
- 230000008569 process Effects 0.000 claims description 17
- 238000004590 computer program Methods 0.000 claims description 6
- 230000004048 modification Effects 0.000 claims description 5
- 238000012986 modification Methods 0.000 claims description 5
- 238000005070 sampling Methods 0.000 claims description 5
- 230000003068 static effect Effects 0.000 claims description 4
- 230000010365 information processing Effects 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 230000007547 defect Effects 0.000 description 3
- 230000001133 acceleration Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
Abstract
The invention provides a dispatching optimization method and a dispatching optimization system for a distributed virtualization client operating system, wherein a physical machine CPU core is distributed to a virtual machine vCPU, the virtual machine vCPU and a node where the physical machine CPU is located are bound, and information of the virtual machine vCPU is transmitted to the client operating system; modifying a scheduling strategy in a client operating system according to vCPU information of the virtual machine, and directly completing the vCPU core allocation of the client operating system of the computing task through the client operating system; and distributing a group of computing tasks with frequent information interaction to the same virtual machine node through the modified scheduling strategy, so as to realize the optimization of the scheduling strategy of the client operating system of the distributed virtualization. Meanwhile, a corresponding terminal and medium are provided. According to the invention, the scheduling strategy of the client operating system is modified, and the computing tasks with frequent information interaction are distributed to the same node, so that the purposes of reducing the information interaction between nodes and the information interaction cost between nodes are achieved.
Description
Technical Field
The invention relates to the technical field of computer virtualization and distributed systems, in particular to a method, a system and a terminal for dispatching and optimizing a client operating system of distributed virtualization, and provides a corresponding computer readable storage medium.
Background
Distributed virtualization, as shown in fig. 1, refers to remote memory access between machines through some communication, so as to implement distributed shared memory, CPU and IO resources. Such distributed virtualization may also provide access support for other hardware devices such as GPUs. For example, the giant virtual machine (GiantVM) provides massive computing and I/O resources for a single virtual machine or even multiple virtual machines by abstracting hardware resources on multiple machines, thereby meeting application scenes with extremely high resource and performance requirements.
In existing distributed virtualization, sharing of memory is achieved using I/O of the network. For example, the giant virtual machine is added with a plurality of functional modules based on QEMU-KVM, including IPI forwarding, interrupt forwarding, I/O forwarding, clock synchronization and distributed shared memory modules, and the machines are connected through RDMA network.
But the network I/O itself has a larger overhead than the normal local memory access, in order to make a smaller amortization overhead, the one-time high latency overhead is amortized to more bytes, and the minimum granularity of the most main cross-node memory sharing in the cross-node information interaction or data transmission cannot be too small. In addition, some existing distributed virtualization works, such as a huge virtual machine, utilize a rewritten Page Fault processing mechanism to realize the sharing of the cross-node memory, and the minimum granularity of the cross-node memory sharing is not less than one Page.
The existing distributed virtualized cross-node memory sharing minimum granularity can cause performance loss. The minimum granularity of cross-node shared memory is much larger than CACHELINE, so for a program that is not rewritten, the probability of pseudo sharing is much greater, and a significant performance degradation is caused. Pseudo-sharing refers to the preemption problem that two threads access to, which would otherwise have resulted from two memory addresses not being shared, but which are within the same minimum granularity. In addition, even without many programs that are pseudo-shared, if many are true-shared, virtual machines can run slower than a single machine due to the large overhead of network I/O.
As can be seen from the foregoing, the existing distributed virtualization technology has a fatal problem, and when computing tasks are distributed over multiple physical devices, some computing tasks across nodes have frequent memory sharing or other information interaction, resulting in high overhead.
Disclosure of Invention
The invention provides a distributed virtualization client operating system scheduling optimization method, a system and a terminal aiming at the defects in the prior art, and provides a corresponding computer readable storage medium.
According to one aspect of the present invention, there is provided a method for optimizing guest operating system scheduling for distributed virtualization, including:
Distributing the CPU core of the physical machine to the vCPU of the virtual machine, binding the vCPU of the virtual machine with the node where the CPU of the physical machine is located, and transmitting the information of the vCPU of the virtual machine to a client operating system;
modifying a scheduling strategy in a client operating system according to information of the virtual machine vCPU, and directly completing the allocation of the vCPU core of the client operating system of the computing task through the client operating system;
and distributing a group of computing tasks with frequent information interaction to the same virtual machine node through the modified scheduling strategy, so as to realize the optimization of the scheduling strategy of the client operating system of the distributed virtualization.
Preferably, the binding the virtual machine vCPU with the node where the physical machine CPU is located includes:
static binding is carried out on nodes where the virtual machine vCPU and the physical machine CPU are located;
Or (b)
And temporarily binding the virtual machine vCPU and the node where the physical machine CPU is located.
Preferably, the information of the vCPU of the virtual machine includes:
the node information of each vCPU and the vCPU information contained in each node.
Preferably, the modifying the scheduling policy in the guest operating system according to the information of the vCPU of the virtual machine includes:
according to the information of the vCPU of the virtual machine, placing the computing tasks with frequent information interaction on the same node, and placing the computing tasks with infrequent information interaction on different nodes to form a new scheduling strategy; wherein:
the frequent information interaction means that: the communication cost caused by information interaction is greater than a set threshold value; wherein the communication cost includes: inter-node communication times and inter-node communication traffic.
Preferably, the communication cost caused by the information interaction is acquired through a thread process relation or through a Perf analysis tool to acquire cross-node access memory and information acquisition.
Preferably, the allocation of guest operating system vCPU cores for computing tasks is accomplished directly through the guest operating system, including:
The allocation is accomplished through the use of auxiliary tools, or by modifying the method of the guest operating system kernel.
Preferably, the modified scheduling policy includes any one or more of the following:
-defining all threads of the same process to the same virtual machine node;
-sampling the frequency of information interaction between threads and assigning a minimum number of cross-node shares;
The modified scheduling strategy reduces the information interaction cost of the distributed virtual machine.
According to another aspect of the present invention, there is provided a distributed virtualized guest operating system scheduling optimization system comprising:
The information processing module distributes the CPU core of the physical machine to the vCPU of the virtual machine, binds the vCPU of the virtual machine with the node where the CPU of the physical machine is located, and transmits the information of the vCPU of the virtual machine to the client operating system;
the scheduling policy modification module modifies the scheduling policy in the client operating system according to the information of the virtual machine vCPU, and the allocation of the vCPU core of the client operating system for computing tasks is directly completed through the client operating system;
And the scheduling policy optimization module distributes a group of computing tasks with frequent information interaction to the same virtual machine node through the modified scheduling policy, so as to realize the scheduling policy optimization of the client operating system of the distributed virtualization.
According to a third aspect of the present invention there is provided a terminal comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor when executing the program being operable to perform the method of any one of the preceding claims or to run the system of the preceding claims.
According to a fourth aspect of the present invention there is provided a computer readable storage medium having stored thereon a computer program which when executed by a processor is operable to perform a method of any of the above, or to run a system as described above.
Due to the adoption of the technical scheme, compared with the prior art, the invention has at least one of the following beneficial effects:
According to the distributed virtualized client operating system scheduling optimization method, system, terminal and medium, the scheduling strategy of the client operating system is modified, and the computing tasks with frequent information interaction are distributed to the same node, so that the purposes of reducing information interaction between nodes and the cost of information interaction between nodes are achieved.
The distributed virtualized client operating system scheduling optimization method, system, terminal and medium provided by the invention realize a new virtualized physical resource scheduling realization technology based on virtualization, and the technology is based on a client operating system scheduling strategy, so that better performance of the distributed shared memory is obtained, and the purpose of reducing cross-node communication overhead is achieved.
The method, the system, the terminal and the medium for dispatching and optimizing the distributed virtualized client operating system can reduce the sharing jitter times of the distributed virtualized memory, bind the frequently-information-interaction computing tasks to the same node aiming at different application scenes and application limitations, and greatly improve the performance of the distributed virtualization.
The distributed virtualized client operating system scheduling optimization method, system, terminal and medium provided by the invention modify the scheduling strategy, namely the vCPU scheduling strategy, in the client operating system for scheduling the CPUs of different nodes, thereby realizing the function of arranging the computing tasks which are possibly frequently shared to the same node as much as possible.
Drawings
Other features, objects and advantages of the present invention will become more apparent upon reading of the detailed description of non-limiting embodiments, given with reference to the accompanying drawings in which:
fig. 1 is a schematic diagram of a distributed virtualization technology in the prior art.
FIG. 2 is a flow chart of a method for optimizing the distributed virtualized guest operating system scheduling in accordance with an embodiment of the present invention.
FIG. 3 is a schematic diagram of a method for optimizing the distributed virtualized guest operating system scheduling in accordance with a preferred embodiment of the present invention.
FIG. 4 is a block diagram illustrating a distributed virtualized guest operating system scheduling system according to one embodiment of the invention.
Detailed Description
The following describes embodiments of the present invention in detail: the embodiment is implemented on the premise of the technical scheme of the invention, and detailed implementation modes and specific operation processes are given. It should be noted that variations and modifications can be made by those skilled in the art without departing from the spirit of the invention, which falls within the scope of the invention.
FIG. 2 is a flowchart illustrating a method for optimizing a distributed virtualized guest operating system schedule according to an embodiment of the present invention.
As shown in fig. 2, the method for optimizing the scheduling of the client operating system for distributed virtualization provided in this embodiment may include the following steps:
s100, distributing a CPU core of a physical machine to a vCPU of a virtual machine, binding the vCPU of the virtual machine and a node where the CPU of the physical machine is located, and transmitting information of the vCPU of the virtual machine to a client operating system;
S200, modifying a scheduling strategy in a client operating system according to information of the virtual machine vCPU, and directly completing the allocation of the vCPU core of the client operating system of the computing task through the client operating system;
And S300, distributing a group of computing tasks with frequent information interaction to the same virtual machine node through the modified scheduling strategy, and realizing the scheduling strategy optimization of the client operating system of the distributed virtualization.
In S100 of this embodiment, as a preferred embodiment, binding the virtual machine vCPU with the node where the physical machine CPU is located includes:
static binding is carried out on nodes where the virtual machine vCPU and the physical machine CPU are located;
Or (b)
And temporarily binding the virtual machine vCPU and the node where the physical machine CPU is located.
In S100 of this embodiment, as a preferred embodiment, information of the vCPU of the virtual machine includes:
the node information of each vCPU and the vCPU information contained in each node.
In S200 of this embodiment, as a preferred embodiment, modifying the scheduling policy in the guest operating system according to the information of the vCPU of the virtual machine includes:
according to the information of the vCPU of the virtual machine, placing the computing tasks with frequent information interaction on the same node, and placing the computing tasks with infrequent information interaction on different nodes to form a new scheduling strategy; wherein:
frequent information interaction means: the communication cost caused by information interaction is greater than a set threshold value; wherein, the communication cost includes: inter-node communication times and inter-node communication traffic.
Further, as a preferred embodiment, the communication cost caused by the information interaction is obtained through a thread process relationship or through a Perf analysis tool to collect cross-node access memory and information acquisition.
In S200 of this embodiment, as a preferred embodiment, the allocation of the guest operating system vCPU core for the computing task is accomplished directly by the guest operating system, including:
The allocation is accomplished through the use of auxiliary tools, or by modifying the method of the guest operating system kernel.
In S300 of this embodiment, as a preferred embodiment, the modified scheduling policy includes any one or more of the following:
-defining all threads of the same process to the same virtual machine node;
-sampling the frequency of information interaction between threads and assigning a minimum number of cross-node shares;
The modified scheduling strategy reduces the information interaction cost of the distributed virtual machine.
FIG. 3 is a schematic diagram of a method for optimizing distributed virtualized guest operating system scheduling according to a preferred embodiment of the present invention.
As shown in fig. 3, the method for optimizing the scheduling of the client operating system for distributed virtualization provided in this embodiment may include: the physical machine CPU core is distributed to the virtual machine vCPU and the guest operating system vCPU core.
Wherein:
the method for distributing the CPU cores of the physical machine to the vCPU of the virtual machine is as follows:
The CPU core of the physical machine is distributed to the vCPU, the virtual machine vCPU and the node where the physical machine CPU is located are subjected to static binding or temporary binding, and the information of the virtual machine vCPU is transmitted to a client operating system. Further allocation of vCPU cores is performed by the guest operating system.
The guest operating system vCPU core allocation method is as follows:
The scheduling policy in the guest operating system is modified. The allocation of the guest operating system vCPU cores for the computing tasks is accomplished directly by the guest operating system. And distributing a group of computing tasks with frequent information interaction to the same virtual machine node through the modified scheduling strategy.
In one specific application example, the modified scheduling policy is: all threads of the same process are restricted to the same virtual machine node. The scheduling policy of the present invention is not limited to the simple implementation policy listed above.
The method for optimizing the dispatching of the client operating system in the distributed virtualization, which is provided by the preferred embodiment, is designed based on the architecture shown in fig. 3, and is specifically used for classifying the computing task groups according to the information interaction frequency degree:
In the original distributed virtual machine, each VM virtual node has a plurality of vCPUs, and each Host physical node has a plurality of CPUs. The default operating system of the distributed virtual machine uses a load balancing scheduling policy, and on top of the operating system, the scheduler distributes these computing tasks evenly to vCPU in each VM to complete, where vCPU and CPU in Host node are corresponding, which results in many communications between computing tasks (tasks) that would otherwise have many communications being distributed to different Host nodes, resulting in a significant communication time penalty.
The technical scheme provided by the embodiment of the invention is mainly an optimization aiming at the defect. After the defect that the original operating system runs on the distributed virtual machine is found, the embodiment of the invention proposes to modify the scheduling strategy in the operating system to realize acceleration on the distributed virtual machine. The scheduling strategy is that the computation tasks with close communication are distributed to the vCPUs corresponding to the same Host node, the computation tasks with not close communication are distributed to the vCPUs of different Host nodes, the number of cross-node (Host) communication and the total cost of communication time are reduced, and the performance of the distributed virtual machine is optimized according to the scheme.
An interface is reserved for a programmer in a standard operating system, so that the programmer can conveniently modify the scheduling strategy; the interface is used for controlling the scheduling in the operating system to schedule according to the information of the vCPU of the virtual machine, so that the purpose of acceleration is achieved. Of course, besides using this interface, the present invention may use other interfaces to perform "scheduling according to information of vCPU of virtual machine", where these interfaces include cpuset, tasksched, taskset and so on, and these interfaces are all protocols of free software, and in whatever form, all fall within the protection scope of the present invention.
The technical solutions provided in the above embodiments of the present invention are described in further detail below in conjunction with a specific application example. However, it should be noted that the use platform of the technical solution provided in the foregoing embodiment of the present invention is not limited to the following example.
In this specific application example, the specific deployment is a cluster of three general servers, each server being equipped with a network card supporting InfiniBand. The servers are connected to a central InfiniBand switch by optical fibers. The technical scheme provided by the embodiment of the invention is not limited by the types, the configurations and the number of the hosts, and can be extended to any number of hosts with the number more than 1 to form a cluster. The technical scheme provided by the embodiment of the invention is not limited by the network card and the network equipment, and any type of network card and network equipment can be used.
Each server is equipped with ubuntuserver16.04.1lts64bit and GiantVM and is equipped with two CPU cores together, 56 cores and 64GB of memory. Specific developments are based on GiantVM, QEMU2.9.0 and source code versions of Linux kernel 4.8.10 as illustrations, as well as other virtual machines, virtual machine managers and Linux other versions of kernel.
The giant virtual machine has 168 vCPUs, each vCPU corresponds to one physical core, wherein 56 vCPUs run on the physical cores of the machine, and the other 112 vCPUs run on the other two remote servers. The huge virtual machine has 192GB of distributed shared memory, wherein 64GB is local memory, the other 128GB is remote memory, and the remote memory is accessed at high speed through RDMA. Meanwhile, virtual machines own and can use I/O devices such as GPUs, FPGAs, etc. that are located on different computers. The technical scheme provided by the embodiment of the invention is not limited by the number of vCPUs, the memory size, the I/O equipment and the communication protocol, and other protocols except RDMA can be optimized by adopting the technical scheme provided by the embodiment of the invention.
The guest operating system is a modified ubuntuxerver 16.04.1lts64bit, where the scheduling policy is a policy for "same process assigned to same virtual node". The technical scheme provided by the embodiment of the invention is not limited by the client operating system, and any other client operating system can be rewritten by similar means. The technical scheme provided by the embodiment of the invention is not limited by the scheduling strategy, and the invention can realize the optimization of information interaction by using different specific strategies.
The technical solution provided in the above embodiment of the present invention includes, but is not limited to, a scheduling policy that "the same process is bound to the same virtual node", where the scheduling policy can accelerate the rationality of distributed virtualization is that: if the guest operating system randomly distributes the vCPU, the underlying virtual machine cannot group the mixed computing tasks onto the CPU of the physical machine according to whether there is a large number of shares; the guest operating system, if it employs the allocation of the same process at the same virtual node, can ensure that the same process does not cross nodes on the physical machine. In this way, inter-thread communications are all at the same node. Because of the different address spaces, the processes do not have direct shared memory originally. The minimum granularity of the shared memory such as MMap among the processes is a page, and the information interaction is carried out among the processes according to the needs, and unless the information interaction is purposely carried out, the condition of pseudo sharing hardly exists.
FIG. 4 is a schematic diagram illustrating a distributed virtualized guest operating system scheduling optimization system according to an embodiment of the present invention.
As shown in fig. 4, the client operating system scheduling optimization system for distributed virtualization provided in this embodiment may include the following modules:
the information processing module distributes the CPU core of the physical machine to the vCPU of the virtual machine, binds the vCPU of the virtual machine with the node where the CPU of the physical machine is located, and transmits the information of the vCPU of the virtual machine to the client operating system;
the scheduling policy modification module modifies the scheduling policy in the client operating system according to the information of the virtual machine vCPU, and the allocation of the vCPU core of the client operating system for computing tasks is directly completed through the client operating system;
And the scheduling policy optimization module distributes a group of computing tasks with frequent information interaction to the same virtual machine node through the modified scheduling policy, so as to realize the scheduling policy optimization of the client operating system of the distributed virtualization.
An embodiment of the present invention provides a terminal including a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor is operable to perform the method of any of the above embodiments of the present invention when the program is executed by the processor.
An embodiment of the present invention provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, is operable to perform a method according to any of the above embodiments.
According to the distributed virtualized client operating system scheduling optimization method, system, terminal and medium, the scheduling strategy of the client operating system is modified, and the computing tasks with frequent information interaction are distributed to the same node, so that the purposes of reducing the information interaction among the nodes and the cost of the information interaction among the nodes are achieved; based on virtualization, a new technology for realizing virtualized physical resource scheduling is realized, and the technology is based on a client operating system scheduling strategy, so that a distributed shared memory obtains better performance, and the cross-node communication overhead is reduced; the information interaction jitter frequency of the distributed virtualization can be reduced, and the computing tasks with frequent information interaction are bound to the same node aiming at different application scenes and application limits, so that the performance of the distributed virtualization is greatly improved; the scheduling strategy, i.e. the vCPU scheduling strategy, is modified in the client operating system and is used for scheduling the CPUs of different nodes, so that the function of arranging the computing tasks which are possibly frequently shared to the same node is realized.
Those skilled in the art will appreciate that the invention provides a system and its individual devices that can be implemented entirely by logic programming of method steps, in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers, etc., in addition to the system and its individual devices being implemented in pure computer readable program code. Therefore, the system and various devices thereof provided by the present invention may be considered as a hardware component, and the devices included therein for implementing various functions may also be considered as structures within the hardware component; means for achieving the various functions may also be considered as being either a software module that implements the method or a structure within a hardware component.
The foregoing describes specific embodiments of the present invention. It is to be understood that the invention is not limited to the particular embodiments described above, and that various changes and modifications may be made by one skilled in the art within the scope of the claims without affecting the spirit of the invention.
Claims (8)
1. A method for optimizing the scheduling of a distributed virtualized guest operating system, comprising:
Distributing the CPU core of the physical machine to the vCPU of the virtual machine, binding the vCPU of the virtual machine with the node where the CPU of the physical machine is located, and transmitting the information of the vCPU of the virtual machine to a client operating system;
modifying a scheduling strategy in a client operating system according to information of the virtual machine vCPU, and directly completing the allocation of the vCPU core of the client operating system of the computing task through the client operating system;
Distributing a group of computing tasks with frequent information interaction to the same virtual machine node through the modified scheduling strategy, so as to realize the optimization of the scheduling strategy of the client operating system of the distributed virtualization;
The modifying the scheduling policy in the guest operating system according to the information of the virtual machine vCPU includes:
according to the information of the vCPU of the virtual machine, placing the computing tasks with frequent information interaction on the same node, and placing the computing tasks with infrequent information interaction on different nodes to form a new scheduling strategy; wherein:
The frequent information interaction means that: the communication cost caused by information interaction is greater than a set threshold value; wherein the communication cost includes: the number of inter-node communication times and the inter-node communication traffic;
The modified scheduling policy includes any one or more of the following:
-defining all threads of the same process to the same virtual machine node;
-sampling the frequency of information interaction between threads and assigning a minimum number of cross-node shares;
The modified scheduling strategy reduces the information interaction cost of the distributed virtual machine.
2. The method for optimizing the scheduling of a guest operating system of distributed virtualization according to claim 1, wherein binding the virtual machine vCPU with the node where the physical machine CPU is located comprises:
static binding is carried out on nodes where the virtual machine vCPU and the physical machine CPU are located;
Or temporarily binding the virtual machine vCPU and the node where the physical machine CPU is located.
3. The method for optimizing guest operating system scheduling of distributed virtualization according to claim 1, wherein the information of the virtual machine vCPU includes:
the node information of each vCPU and the vCPU information contained in each node.
4. The method for optimizing the scheduling of a distributed virtualized guest operating system according to claim 1, wherein the communication cost caused by the information interaction is acquired through a thread process relation or through a Perf analysis tool to acquire cross-node access memory and information acquisition.
5. The method of claim 1, wherein assigning guest operating system vCPU cores for computing tasks directly by a guest operating system, comprises:
the allocation is accomplished by using auxiliary tools, or by modifying the method of the guest operating system kernel;
The modified scheduling policy includes any one or more of the following:
-defining all threads of the same process to the same virtual machine node;
-sampling the frequency of information interaction between threads and assigning a minimum number of cross-node shares;
The modified scheduling strategy reduces the information interaction cost of the distributed virtual machine.
6. A distributed virtualized guest operating system scheduling optimization system comprising:
The information processing module distributes the CPU core of the physical machine to the vCPU of the virtual machine, binds the vCPU of the virtual machine with the node where the CPU of the physical machine is located, and transmits the information of the vCPU of the virtual machine to the client operating system;
the scheduling policy modification module modifies the scheduling policy in the client operating system according to the information of the virtual machine vCPU, and the allocation of the vCPU core of the client operating system for computing tasks is directly completed through the client operating system;
The scheduling policy optimization module distributes a group of computing tasks with frequent information interaction to the same virtual machine node through the modified scheduling policy, so as to realize the scheduling policy optimization of the client operating system of the distributed virtualization;
The modifying the scheduling policy in the guest operating system according to the information of the virtual machine vCPU includes:
according to the information of the vCPU of the virtual machine, placing the computing tasks with frequent information interaction on the same node, and placing the computing tasks with infrequent information interaction on different nodes to form a new scheduling strategy; wherein:
The frequent information interaction means that: the communication cost caused by information interaction is greater than a set threshold value; wherein the communication cost includes: the number of inter-node communication times and the inter-node communication traffic;
The modified scheduling policy includes any one or more of the following:
-defining all threads of the same process to the same virtual machine node;
-sampling the frequency of information interaction between threads and assigning a minimum number of cross-node shares;
The modified scheduling strategy reduces the information interaction cost of the distributed virtual machine.
7. A terminal comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor is operable to perform the method of any one of claims 1-5 or to run the system of claim 6 when the program is executed by the processor.
8. A computer readable storage medium having stored thereon a computer program, which when executed by a processor is operative to perform the method of any one of claims 1-5 or to run the system of claim 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111508295.XA CN114281529B (en) | 2021-12-10 | Method, system and terminal for dispatching optimization of distributed virtualized client operating system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111508295.XA CN114281529B (en) | 2021-12-10 | Method, system and terminal for dispatching optimization of distributed virtualized client operating system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114281529A CN114281529A (en) | 2022-04-05 |
| CN114281529B true CN114281529B (en) | 2024-06-04 |
Family
ID=
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108932154A (en) * | 2018-07-23 | 2018-12-04 | 上海交通大学 | A kind of distributed virtual machine manager |
| CN110704195A (en) * | 2019-10-14 | 2020-01-17 | 腾讯云计算（北京）有限责任公司 | CPU adjusting method, server and computer readable storage medium |
| CN111143025A (en) * | 2019-11-22 | 2020-05-12 | 中国船舶工业***工程研究院 | Method for real-time virtual machine instance management |
| CN112256383A (en) * | 2019-07-22 | 2021-01-22 | 深信服科技股份有限公司 | Method, device, equipment and medium for adjusting CPU core number of virtual machine |
| CN112667363A (en) * | 2021-01-05 | 2021-04-16 | 浪潮云信息技术股份公司 | Method and device for simulating cloud physical host by using virtual machine based on cloud platform |
| CN113032154A (en) * | 2021-04-19 | 2021-06-25 | 深信服科技股份有限公司 | Virtual CPU scheduling method and device, electronic equipment and storage medium |
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108932154A (en) * | 2018-07-23 | 2018-12-04 | 上海交通大学 | A kind of distributed virtual machine manager |
| CN112256383A (en) * | 2019-07-22 | 2021-01-22 | 深信服科技股份有限公司 | Method, device, equipment and medium for adjusting CPU core number of virtual machine |
| CN110704195A (en) * | 2019-10-14 | 2020-01-17 | 腾讯云计算（北京）有限责任公司 | CPU adjusting method, server and computer readable storage medium |
| CN111143025A (en) * | 2019-11-22 | 2020-05-12 | 中国船舶工业***工程研究院 | Method for real-time virtual machine instance management |
| CN112667363A (en) * | 2021-01-05 | 2021-04-16 | 浪潮云信息技术股份公司 | Method and device for simulating cloud physical host by using virtual machine based on cloud platform |
| CN113032154A (en) * | 2021-04-19 | 2021-06-25 | 深信服科技股份有限公司 | Virtual CPU scheduling method and device, electronic equipment and storage medium |
Non-Patent Citations (3)
| Title |
|---|
| "CIVSched: Communication-aware Inter-VM Scheduling in Virtual Machine Monitor Based on the Process";Bei Guan;《2013 13th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing》;20130624;第597-604页 * |
| "GiantVM: A Novel Distributed Hypervisor for Resource Aggregation with DSM-aware Optimizations";Xingguo Jia;《ACM Transactions on Architecture and Code Optimization》;20220307;第19卷(第2期);第1-27页 * |
| "异构环境下虚拟资源一体化管理及分配机制的研究与应用";姚栋杰;《中国优秀硕士学位论文全文数据库 信息科技辑》;20181215(第2018年12期);第I139-128页 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11068293B2 (en) | Parallel hardware hypervisor for virtualizing application-specific supercomputers | |
| Ibanez et al. | The nanopu: A nanosecond network stack for datacenters | |
| US11010053B2 (en) | Memory-access-resource management | |
| US9519795B2 (en) | Interconnect partition binding API, allocation and management of application-specific partitions | |
| US8201170B2 (en) | Operating systems are executed on common program and interrupt service routine of low priority OS is modified to response to interrupts from common program only | |
| US20160378545A1 (en) | Methods and architecture for enhanced computer performance | |
| KR102103596B1 (en) | A computer cluster arragement for processing a computation task and method for operation thereof | |
| US20050251806A1 (en) | Enhancement of real-time operating system functionality using a hypervisor | |
| US20090327610A1 (en) | Method and System for Conducting Intensive Multitask and Multiflow Calculation in Real-Time | |
| US20070150895A1 (en) | Methods and apparatus for multi-core processing with dedicated thread management | |
| KR20060071307A (en) | Systems and methods for exposing processor topology for virtual machines | |
| Cheng et al. | vScale: Automatic and efficient processor scaling for SMP virtual machines | |
| WO2012016439A1 (en) | Method, device and equipment for service management | |
| US20110219373A1 (en) | Virtual machine management apparatus and virtualization method for virtualization-supporting terminal platform | |
| Ahmadinia et al. | Task scheduling for heterogeneous reconfigurable computers | |
| CN114168271B (en) | Task scheduling method, electronic device and storage medium | |
| CN112306669A (en) | Task processing method and device based on multi-core system | |
| Hetherington et al. | Edge: Event-driven gpu execution | |
| CN114281529B (en) | Method, system and terminal for dispatching optimization of distributed virtualized client operating system | |
| CN114281529A (en) | Distributed virtualized client operating system scheduling optimization method, system and terminal | |
| CN113556264A (en) | Real-time cloud platform management monitoring system | |
| US9619277B2 (en) | Computer with plurality of processors sharing process queue, and process dispatch processing method | |
| CN107273188B (en) | Virtual machine Central Processing Unit (CPU) binding method and device | |
| Uhlig | The mechanics of in-kernel synchronization for a scalable microkernel | |
| CN114816678B (en) | Virtual machine scheduling method, system, equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |