KR20230108059A - Apparatus and method for distributed allocation of memory resources based on priority - Google Patents

Abstract

A resource allocation method and apparatus for a data processing server are disclosed. The resource allocation method includes receiving a resource allocation request for service execution from a client device, determining whether the service is a delay-sensitive service in response to the resource allocation request, and determining whether the service is a delay-sensitive service. , determining whether there are system resources available in the current blade server, if there are system resources available in the current blade server, allocating system resources of the current blade server to the client device, and If there are no system resources, determining whether there are system resources available in the rack server; and if there are system resources available in the rack server, system resources of the rack server are assigned to client devices based on priority. It includes allocating

Description

Method and apparatus for distributing and allocating memory resources based on priority

The disclosure below relates to a technique for distributing and allocating memory resources based on priorities.

Unlike server-centric data centers where resources such as processors and memory are fixed in one server, resource-centric data centers organize and operate each compute and memory resource as a pool Resource-centric data centers are evolving. Research on an optical transmission technology capable of providing ultra-low latency and ultra-high capacity during the operation of a resource-oriented data center is also being actively conducted.

A resource allocation method for a data processing server according to an embodiment includes receiving a resource allocation request for service execution from a client device; determining whether the service is a delay-sensitive service in response to the resource allocation request; determining whether system resources are currently available in the blade server if the service is the delay sensitive service; allocating system resources of the current blade server to the client device if there are system resources available in the current blade server; and determining whether there are available system resources in a rack server when there are no available system resources in the current blade server. and allocating system resources of the rack server to the client device based on priority, when there are system resources available in the rack server.

The resource allocation method may include determining whether there are system resources available in the entire rack server when the service is not the delay-sensitive service; and allocating system resources of the rack server to the client device when system resources are available in all rack servers.

The step of determining whether there are system resources available in the rack server may further include transmitting a resource allocation failure notification to the client device when there are no system resources available in the rack server.

Allocating system resources of the rack server to the client device may include updating the system resource allocation information and transmitting the system resource allocation information to the client device and the rack server when the system resource allocation is completed. may further include.

The priority may be determined based on information including memory resource capacity, computing resource capacity, service type, service level agreement, and quality of service of the resource allocation request.

The priority may be determined based on information further including information on a storage device capacity, an accelerator capacity, and the number of computing resources of the resource allocation request.

The system resources may include memory resources and computing resources.

The computing resource module performs access to the memory resource module based on frame data for memory access, and the frame data includes information for distinguishing whether the service is a memory-focused service or a computing-focused service, and priority information. can include

The frame data may further include service level agreement information and quality of service information.

An apparatus for allocating resources for a data processing server according to an embodiment includes a memory for storing one or more instructions; And a processor for executing the instruction, when the instruction is executed, the processor receives a resource allocation request for performing a service from a client device, and in response to the resource allocation request, the service is delay sensitive ( delay-sensitive service, and if the service is the delay-sensitive service, determining whether there are system resources available in the current blade server, and if there are system resources available in the current blade server, Allocating system resources of the current blade server to the client device, and determining whether there are available system resources in a rack server when there are no available system resources in the current blade server, and If there are system resources available in the server, system resources of the rack server may be allocated to the client devices based on priority.

1 is a block diagram illustrating a configuration of a data center that distributes and allocates memory resources based on priorities according to an exemplary embodiment.
2 is a block diagram illustrating a configuration within a rack of a data center that distributes and allocates memory resources based on priorities according to an exemplary embodiment.
3 is a flowchart illustrating an operation of an apparatus for distributing and allocating memory resources based on priorities according to an exemplary embodiment.
4 is a flowchart illustrating an operation of an apparatus for distributing and allocating memory resources based on delay-sensitive service availability according to an embodiment.
5 is a diagram for explaining a frame format for memory access of an apparatus for distributing and allocating memory resources based on priorities according to an exemplary embodiment.
6 is a block diagram showing a schematic configuration of an apparatus for distributing and allocating memory resources according to an exemplary embodiment.

Specific structural or functional descriptions of the embodiments are disclosed for illustrative purposes only, and may be changed and implemented in various forms. Therefore, the form actually implemented is not limited only to the specific embodiments disclosed, and the scope of the present specification includes changes, equivalents, or substitutes included in the technical idea described in the embodiments.

Although terms such as first or second may be used to describe various components, such terms should only be construed for the purpose of distinguishing one component from another. For example, a first element may be termed a second element, and similarly, a second element may be termed a first element.

It should be understood that when an element is referred to as being “connected” to another element, it may be directly connected or connected to the other element, but other elements may exist in the middle.

Singular expressions include plural expressions unless the context clearly dictates otherwise. In this specification, terms such as "comprise" or "have" are intended to designate that the described feature, number, step, operation, component, part, or combination thereof exists, but one or more other features or numbers, It should be understood that the presence or addition of steps, operations, components, parts, or combinations thereof is not precluded.

Unless defined otherwise, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by one of ordinary skill in the art. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning in the context of the related art, and unless explicitly defined in this specification, it should not be interpreted in an ideal or excessively formal meaning. don't

Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. In the description with reference to the accompanying drawings, the same reference numerals are given to the same components regardless of reference numerals, and overlapping descriptions thereof will be omitted.

1 is a block diagram illustrating a configuration of a data center that distributes and allocates memory resources based on priorities according to an exemplary embodiment.

Referring to FIG. 1 , a data center 10 may include a gateway 110 , a spine switch 120 , and a plurality of racks 130 . The data center 10 may interconnect with an external network through the gateway 110 , and the spine switch 120 may connect a plurality of racks 130 .

Data center 10 may be a separate data center. The data center 10 has a structure in which computing resources and memory resources are composed of different boards and connected remotely, unlike existing server-oriented data centers in which computing resources and memory resources are locally connected to a single board. can have

The racks 130 constituting the internal network of the data center 10 include an optical line switch 230, a plurality of disaggregated computing (dComp) modules 220, and a plurality of disaggregated memory (dMem) modules 250. can do. In each rack 130 of the data center 10, separated computing and memory resources (eg, dComp module 220, dMem module 240) are connected through an optical circuit switch 230 to efficiently use resources. To this end, a separate resource management and network connection control method may be provided. Embodiments may efficiently manage specific resources and network resources of a computing pool and a memory pool separated by distance from one rack 130 in a disaggregated data center 10 .

2 is a block diagram illustrating a configuration within a rack of a data center that distributes and allocates memory resources based on priorities according to an exemplary embodiment.

Referring to FIG. 2 , the dComp module 220 is a module that performs various computational operations (eg, computational tasks), and includes a CPU (eg, multi-core processor) (not shown) and optical transceivers (not shown). can do. The dComp module 220 may use local memory to perform calculation work, and may use the memory of the dMem module 240 if necessary.

The dMem module 240 is a large-capacity memory device composed of a plurality of memories, and may provide memory access functions (eg, read and write) according to a request of the dComp module 220 . The dMem module 240 is a memory pool shared and used by each dComp module 220, a memory read/write control device (eg, controller, memory controller) (not shown), a plurality of memories (not shown) connected thereto time), and optical transceivers (not shown).

The optical line switch 230 may couple the dComp module 220 with one or more dMem modules 240 . The optical line switch 230 may switch and output optical signals to different output ports according to the wavelengths of input optical signals. Since the optical line switch 230 has a characteristic in which the wavelengths switched according to the input port are circulated, even when wavelength division multiplexed optical signals composed of the same wavelengths are input to different input ports, the same wavelength signals are sent to the same port. It is possible to prevent optical signals from being output, so that optical signals of the same wavelength can be switched without interference (or collision). The optical circuit switch 230 may be implemented as an ultra-high speed optical circuit switch (OCS) to provide low latency. In addition, the optical line switch 230 may be implemented as an arrayed waveguide grating (AWG) switch.

A plurality of dComp modules 220 may be connected to K dMem modules 240 through an optical line switch 230 . In FIG. 2, it is shown that the number of dComp modules 220 is N (eg, a natural number greater than 1) and the number of dMem modules 240 is K (eg, a natural number greater than 1), but is not limited thereto, and N and K are They may be the same or different. The number of dComp modules 220 and dMem modules 240 constituting the connection may be the same or different.

One dComp module 220 may be connected to each dMem module 240 through a separate fixed port. For example, the number of ports of each dComp module 220 is equal to the number of dMem modules 240 connected to the optical line switch 230, and the optical line switch 230 is (N*K) * (N*K) ) switches (e.g. switches of N*K input ports and N*K output ports) may be used. At this time, one dComp module 220 may access one or more dMem modules 240 to use memory resources. In addition, the plurality of dComp modules 220 may access resources (eg, memory) of one dMem module 240 according to the allocated memory range (eg, determined by the allocated memory capacity, start address, and end address). can

The resource manager 210 may collect resource information about the dComp module 220 , resource information about the dMem module 240 , and resource information about the optical line switch 230 (eg, optical switch information). The resource management unit 210 may manage and control network connections and shared resources between resources (eg, computing resources and memory resources) using the collected resource information.

The resource management unit 210 may manage a resource management table using the collected resource information. The resource management table may include a resource information collection and management unit, a resource allocation unit, a transmission unit, and a reception unit. The resource information collection and management unit may include resource information for the dComp module 220, and the resource allocation unit may include resource information for the dMem module 240. The resource information collection and management unit may include optical path setting information between the dComp module 220 and the dMem module 240 .

The resource management unit 210 receives a resource allocation request (eg, including the capacity of computing resources and memory resources) from a client, and includes one or more dComp modules 220 and one or more dMem modules to configure a virtual server according to the resource allocation request. 240) may determine resource allocation information. The resource management unit 210 may generate optical path setting information for optical path connection between one or more dComp modules 220 and one or more dMem modules 240 to constitute a virtual server, and may update a resource management table.

The resource management unit 210 transmits optical path setting information and resource allocation information to one or more dComp modules 220 and one or more dMem modules 240 constituting a virtual server, and the corresponding dComp modules 220 and dMem modules 240 In order to control the optical path of ), a control signal may be transmitted to the optical line switch 230. The control signal may include connection information about optical paths between the corresponding dComp module 220 and the dMem module 240 .

The dComp module 220 requests a read/write (R/W) request (e.g., command) can be sent. At this time, in the case of writing, the dComp module 220 may transmit data together with a write request. The dMem module 240 may process requests received from the dComp module 220 (eg, read/write (R/W) requests) and perform read and write commands on the memory allocated to the dComp module 220. there is. At this time, the dMem module 240 sends commands and memory address information (eg, memory start address and memory end address) to the memory controller included in the dMem module 240 in order to perform memory read/write (R/W). can be conveyed In the case of a read command, the dMem module 240 transmits the data read from the memory to the corresponding dComp module 220 in the case of a read command, and in the case of a write command, writes the received data to the memory and sends it to the corresponding dComp module 220. ack can be sent. When the operation between the dComp module 220 and the dMem module 240 is completed, the resource management unit 210 provides an optical path between corresponding resources (eg, computing resources of the dComp module 220 and memory resources of the dMem module 240). can cancel In FIG. 2 , each of the racks 130 is illustrated as having the same type of modules having the same connection structure, but each of the racks 130 may include different types of modules and each of the modules may be arranged with different connection structures.

3 is a flowchart illustrating an operation of an apparatus for distributing and allocating memory resources based on priorities according to an exemplary embodiment.

Existing server-centered data centers in which resources such as processors and memory are fixed in one server have a problem in that the utilization rate of entire resources is very low as many applications intensively use only processor and memory resources. Recently, research on a resource-oriented data center is being conducted in order to efficiently use data center resources. Using a resource-oriented data sensor allocates distributed computing and memory resources in one rack based on a client's request for resource allocation for services, and manages resources efficiently by managing them in a server unit. You can do it. Hereinafter, a method for sharing distributed computing and memory resources in a resource-oriented data center and distributing and allocating necessary resources based on priority is provided. This allocates distributed resources according to traffic characteristics and priorities, prevents high-priority resources from being intensively allocated to specific memory resources, minimizes performance degradation due to delay, and guarantees minimum delay according to priority. can do.

Referring to FIG. 3 , in step 310, the resource allocation device may collect resource information and create a resource management table based on the collected resource information. The resource allocation unit is a resource management table, which includes the capacity of resources (e.g., computing and memory resources), the number of resources, port and wavelength information, optical switch information (e.g., the number of input/output ports of the optical switch, wavelength information), Information such as network connection (eg, optical path) information of resources and memory resources, availability of available resources, and the like can be managed.

In step 320, the resource allocation device may receive a resource allocation/cancellation request from the client. The resource allocation request may be a request for allocation of computing resources and memory resources constituting a virtual server. The resource release request may be a request from a client device to release an already allocated resource.

In step 325, the resource allocation device may allocate resources in step 330 if there is a resource allocation request according to whether a resource allocation request is made from the client device, and if there is no resource allocation request in step 340, the resource and The optical path can be terminated. In step 325, if there is a resource allocation request, the resource allocation device determines resource allocation information for one or more dComp modules and one or more dMem modules constituting the virtual server based on the resource management table, and the resource management table (eg : compute resource management table, memory resource management table, compute-memory interconnection table), and set the optical path between the dComp module and the dMem module.

In step 340, the resource allocation device may receive a resource release request from the client device. The resource cancellation request may be a cancellation request for a computing resource and a memory resource constituting a virtual server when all tasks of configured virtual servers are terminated.

In step 350, the resource allocation device may update the resource management table. When the resource allocation device receives the resource connection release response message (dR_release_response), it releases the optical path connected between the dComp module and the dMem module, and the resource management table (eg, computing resource management table, memory resource management table, compute-memory interconnection table) can be updated.

The resource allocation device may perform an operation of dynamically additionally allocating memory allocated to a specific dComp module or dynamically releasing a part of memory according to a request of a client. In this case, the memory resource to be additionally allocated or released may be a memory resource of a dMem module to which an optical path is already connected to a specific dComp module.

4 is a flowchart illustrating an operation of an apparatus for distributing and allocating memory resources based on delay-sensitive service availability according to an embodiment.

In step 410, the device for distributing and allocating memory resources may receive resource allocation request information from the client device.

In step 415, the device for distributing and allocating memory resources may determine whether the resource allocation request from the client device is a delay-sensitive service. In this case, if the service is memory-intensive, it may correspond to a delay-sensitive service, and if it is a processor-intensive service, it may not correspond to a delay-sensitive service.

In step 420, the device for distributing and allocating memory resources determines whether there are available processor resources and memory resources in the same blade server when the resource allocation request from the client device in step 410 is a delay-sensitive service. can

In step 435, the apparatus for distributing and allocating memory resources may determine whether processor resources available in the rack server exist when there are no available processor resources and memory resources in the same blade server in step 420. . If there is no processor resource available in the rack server, the device may send a resource allocation failure notification at step 430 .

In step 455, the apparatus for distributing and allocating memory resources may determine whether there are available memory resources in the rack server when there are processor resources available in the rack server in step 435. If there are no available memory resources in the rack server, the device may send a resource allocation failure notification at step 430 .

In step 440, the apparatus for distributing and allocating memory resources may allocate resources in consideration of priority when available memory resources exist in the rack server in step 455. In step 440, the device for distributing and allocating memory resources considers the priority of the memory and the usage information according to the service type, and with respect to the processor resource and the priority K, the priority of each dMem module is higher than K or equal to K. It is possible to preferentially select and allocate a memory having a low number through the use information of . When allocating processor resources and memory resources in different blade servers in step 440, it is possible to determine whether memory is available by considering whether or not bandwidth is available for the processor resource and the memory resource network path. In step 450, resource allocation information is updated, and in step 460, resource allocation information and control information may be transmitted to a resource management unit.

In step 425, the device for distributing and allocating memory resources may determine whether there are processor resources available in the rack server when the resource allocation request from the client device in step 415 is not a delay-sensitive service. . If there is no processor resource available in the rack server, the device may send a resource allocation failure notification at step 430 .

In step 445, the apparatus for distributing and allocating memory resources may determine whether there are available memory resources in the rack server when there are processor resources available in the rack server in step 425. If there are no available memory resources in the rack server, the device may send a resource allocation failure notification at step 430 . If there are memory resources available within the rack server, the device may allocate the resources at step 440 .

5 is a diagram for explaining a frame format for memory access of an apparatus for distributing and allocating memory resources based on priorities according to an exemplary embodiment.

Referring to FIG. 5 , a memory access frame format 500 used between dComp and dMem modules to access memory resources based on priority in a rack server can be known. The memory access frame format 500 is largely composed of a preamble 505, an SFD 510, a memory access frame header 515, and a payload 520 for memory read/write. The preamble 505 may include information for synchronization for reading information from the memory access frame format 500 . The SFD 510 may include maintenance function driver information of the memory access frame format 500 . The payload 520 may include data information transmitted in use of the memory access frame format 500 .

The memory access frame header 515 is a field for transmitting control information for memory access, and includes BID (525), DCID (530), SCID (535), TAG (540), TYPE (545), and R/W (550). ), M/C (555), PRI (560), S (565), E (570), ACK (575), REQ_SIZE (580), and START_ADDR (585) fields. BID 525, DCID 530, and SCID 535 represent blade IDs, destination component IDs, and source component IDs, respectively, and components represent system resources such as dComp and dMem. The TAG 540 field can be used to distinguish requests in each connection, and the TYPE 545 field can have a value of 0 for a request and 1 for a response to distinguish a request from a response. The R/W (550), S (565), E (570), and ACK (575) fields can be used to distinguish memory read/write commands, start frame, last frame, and notification of memory write, respectively.

In order to support resource allocation according to service type or priority in the dMem module, the M/C (555) field and the PRI (560) field may be used. Here, the M/C (555) field can be used to distinguish a memory-focused service from a CPU-focused service, and the PRI (560) field can be used to indicate priority information.

In addition, the REQ_SIZE (580) field and the START_ADDR (585) field can be used for data size for memory read/write and memory start address information. According to an embodiment, information for indicating service characteristics or QoS may be separately added, and the size of each field may be changed according to an applied scale or implementation method.

6 is a block diagram showing a schematic configuration of an apparatus for distributing and allocating memory resources according to an exemplary embodiment.

The resource allocation device 600 may include a communication interface 620, a memory 630, and a processor 640. Each component 620 to 640 of the resource allocation device 600 may communicate with each other through the bus 610 . The communication interface 620 may be a disaggregated management interface (DMI).

The memory 630 may store instructions (eg, programs) executable by the processor 640 . For example, the instructions may include instructions for executing an operation of the processor 640 and/or an operation of each component of the processor 640 .

The processor 640 may process data stored in the memory 630 . Processor 640 may execute computer readable code (eg, software) stored in memory 630 and instructions invoked by processor 640 .

The processor 640 may be a hardware-implemented data processing device having a circuit having a physical structure for executing desired operations. For example, desired operations may include codes or instructions included in a program.

For example, a data processing unit implemented in hardware includes a microprocessor, a central processing unit, a processor core, a multi-core processor, and a multiprocessor. , Application-Specific Integrated Circuit (ASIC), and Field Programmable Gate Array (FPGA).

An operation performed by the processor 640 may be substantially the same as that of the apparatus for distributing and allocating memory resources described with reference to FIGS. 1 to 5 . Each component of the resource allocation device 600 shown in FIG. 6 (eg, the communication interface 620 and the memory 630) may be executed by the processor 640. Accordingly, detailed descriptions will be omitted.

The embodiments described above may be implemented as hardware components, software components, and/or a combination of hardware components and software components. For example, the devices, methods and components described in the embodiments may include, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate (FPGA). array), programmable logic units (PLUs), microprocessors, or any other device capable of executing and responding to instructions. The processing device may execute an operating system (OS) and software applications running on the operating system. A processing device may also access, store, manipulate, process, and generate data in response to execution of software. For convenience of understanding, there are cases in which one processing device is used, but those skilled in the art will understand that the processing device includes a plurality of processing elements and/or a plurality of types of processing elements. It can be seen that it can include. For example, a processing device may include a plurality of processors or a processor and a controller. Other processing configurations are also possible, such as parallel processors.

Software may include a computer program, code, instructions, or a combination of one or more of the foregoing, which configures a processing device to operate as desired or processes independently or collectively. You can command the device. Software and/or data may be any tangible machine, component, physical device, virtual equipment, computer storage medium or device, intended to be interpreted by or provide instructions or data to a processing device. , or may be permanently or temporarily embodied in a transmitted signal wave. Software may be distributed on networked computer systems and stored or executed in a distributed manner. Software and data may be stored on computer readable media.

The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer readable medium. The computer readable medium may include program instructions, data files, data structures, etc. alone or in combination, and the program instructions recorded on the medium may be specially designed and configured for the embodiment or may be known and usable to those skilled in the art of computer software. may be Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks and magnetic tapes, optical media such as CD-ROMs and DVDs, and magnetic media such as floptical disks. - includes hardware devices specially configured to store and execute program instructions, such as magneto-optical media, and ROM, RAM, flash memory, and the like. Examples of program instructions include high-level language codes that can be executed by a computer using an interpreter, as well as machine language codes such as those produced by a compiler.

The hardware device described above may be configured to operate as one or a plurality of software modules to perform the operations of the embodiments, and vice versa.

As described above, although the embodiments have been described with limited drawings, those skilled in the art can apply various technical modifications and variations based on this. For example, the described techniques may be performed in an order different from the method described, and/or components of the described system, structure, device, circuit, etc. may be combined or combined in a different form than the method described, or other components may be used. Or even if it is replaced or substituted by equivalents, appropriate results can be achieved.

Therefore, other implementations, other embodiments, and equivalents of the claims are within the scope of the following claims.

Claims (18)

A resource allocation method for a data processing server,
Receiving a resource allocation request for service execution from a client device;
determining whether the service is a delay-sensitive service in response to the resource allocation request;
determining whether system resources are currently available in the blade server if the service is the delay sensitive service;
allocating system resources of the current blade server to the client device if there are system resources available in the current blade server; and
determining whether there are available system resources in a rack server when there are no available system resources in the current blade server; and
Allocating system resources of the rack server to the client device based on priority if there are system resources available in the rack server.
A resource allocation method that includes.
According to claim 1,
If the service is not the delay sensitive service, determining whether there are system resources available in the entire rack server; and
Allocating system resources of the rack server to the client device if there are system resources available in the entire rack server.
A resource allocation method further comprising a.
According to claim 1,
The step of determining whether there are system resources available in the rack server,
Sending a resource allocation failure notification to the client device when there are no system resources available in the rack server.
A resource allocation method further comprising a.
According to claim 1,
The step of allocating system resources of the rack server to the client device,
When the system resource allocation is complete, updating the system resource allocation information and transmitting the system resource allocation information to the client device and the rack server.
A resource allocation method further comprising a.
According to claim 1,
The priority is,
Which is determined based on information including memory resource capacity, computing resource capacity, service type, service level agreement, and quality of service of the resource allocation request,
How resources are allocated.
According to claim 5,
The priority is,
Which is determined based on information further including information on the storage device capacity, accelerator capacity, and number of computing resources of the resource allocation request,
How resources are allocated.
According to claim 1,
The system resource,
including memory resources and computing resources,
How resources are allocated.
According to claim 7,
The module for the computing resource,
Accessing the memory resource module based on frame data for memory access;
The frame data,
Including information for distinguishing whether the service is a memory-oriented service or a computing-oriented service, and priority information,
How resources are allocated.
According to claim 8,
The frame data,
Further comprising service level agreement information, and quality of service information,
How resources are allocated.
A resource allocation device for a data processing server,
a memory that stores one or more instructions; and
A processor to execute the instruction
including,
When the instruction is executed, the processor:
Receives a resource allocation request for service execution from a client device, determines whether the service is a delay-sensitive service in response to the resource allocation request, and if the service is the delay-sensitive service, currently Determine whether there are system resources available in the blade server, and if there are system resources available in the current blade server, allocate system resources of the current blade server to the client device, and allocate the system resources in the current blade server to the client device. If there are no available system resources, it is determined whether there are system resources available in the rack server, and if there are system resources available in the rack server, the client device is assigned to the rack server based on priority. of system resources,
resource allocation device.
According to claim 10,
the processor,
If the service is not the delay-sensitive service, it is determined whether there are system resources available in the entire rack server, and if there are system resources available in the entire rack server, the system of the rack server is assigned to the client device. allocating more resources
resource allocation device.
According to claim 10,
the processor,
Further sending a resource allocation failure notification to the client device when there are no available system resources in the rack server,
resource allocation device.
According to claim 10,
the processor,
When the system resource allocation is complete, updating the system resource allocation information and further transmitting the system resource allocation information to the client device and the rack server.
resource allocation device.
According to claim 10,
the processor,
Determining the priority based on information including memory resource capacity, computing resource capacity, service type, service level agreement, and quality of service of the resource allocation request,
resource allocation device.
According to claim 14,
the processor,
Determining the priority based on information further including information on the number of storage device capacity, accelerator capacity, and computing resources of the resource allocation request,
resource allocation device.
According to claim 10,
The system resource,
including memory resources and computing resources,
resource allocation device.
According to claim 16,
The module for the computing resource,
Accessing the memory resource module based on frame data for memory access;
The frame data,
Including information for distinguishing whether the service is a memory-oriented service or a computing-oriented service, and priority information,
resource allocation device.
According to claim 17,
The frame data,
Further comprising service level agreement information and communication quality of service information,
resource allocation device.

Images