WO2017181853A1

WO2017181853A1 - Method, device, and system for dynamically allocating memory

Info

Publication number: WO2017181853A1
Application number: PCT/CN2017/079715
Authority: WO
Inventors: 牛功彪; 张夏涛; 邹巍; 张文涛; 蔡进; 李舒
Original assignee: 阿里巴巴集团控股有限公司
Priority date: 2016-04-20
Filing date: 2017-04-07
Publication date: 2017-10-26
Also published as: TWI795354B; CN107305506A; TW201738754A

Abstract

A method, device, and system for dynamically allocating memory. The method comprises: receiving a memory allocation request from a server (S110); determining, according to the memory allocation request, and on the basis of a memory pool comprising a plurality of memory chips driven by a PCIE apparatus, whether a total available memory size of one or more memory chips satisfies a requested memory size (S120); and if so, allocating to the server the requested memory size (S130). The method, device, and system can realize dynamic memory allocation for a server.

Description

Method, device and system for dynamically allocating memory

The present application claims the priority of the Chinese Patent Application Serial No. No. No. No. No. No. No. No. No. No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No No

Technical field

The present invention relates to the field of computer technologies, and in particular, to a method, an apparatus, and a system for dynamically allocating memory.

Background technique

In computer systems, memory is one of the key factors determining the performance of the whole machine. In a server system, DIMMs (Dual-Inline-Memory-Modules) are often used as the memory. The granules in the DIMM memory are packaged in a DIP (Dual Inline Package) package. The early memory particles were soldered directly to the motherboard so that if a piece of memory fails, the entire motherboard is scrapped. Later, a memory grain slot appeared on the motherboard, so that the memory particles could be replaced. The more commonly used memory particles are DRAM (Dynamic Random Access Memory) chips. A DIMM includes one or more DRAM chips on a small integrated circuit board that can be directly connected to a computer motherboard using pins on the board.

DIMMs are often used in server systems. Due to the large capacity limitations of DIMMs, from 8G, 16G, 32G, 64G to 128G, the single capacity is small, and the lack of flexibility relative to rapid business changes. The current practice is to insert 16 DIMMs in a single server (single machine), or extend the DIMM through the RAZER card to achieve the purpose of expanding the memory capacity. However, since the DIMM is limited by the DIMM CHANNEL of the CPU, there is a limitation in capacity, which results in a server of different specifications due to the capacity requirement, which brings management inconvenience and overhead.

Summary of the invention

One of the technical problems solved by the present invention is to provide a method, device and system for dynamically allocating memory.

According to an embodiment of an aspect of the present invention, a method for dynamically allocating memory includes: receiving a memory allocation request of at least one server; and based on the memory allocation request, based on being driven by a plurality of bus interface standard devices A memory pool composed of memory particles, determining whether the memory pool has a memory sum of one or more free memory particles satisfying the requested memory size; if so, allocating the requested memory to the server.

Preferably, the memory particles comprise DRAM particles, and the method further comprises: DRAM of DRAM particles The interface is converted to a PCIE interface.

Preferably, converting the DRAM interface of the DRAM particle into a PCIE interface comprises: expanding a capacity of the DRAM interface through a memory buffer; connecting the input of the memory controller to the DRAM particle, and performing a double rate DDR memory process on the memory controller to The conversion logic of the PCIE process makes the output of the memory controller a PCIE interface.

Preferably, driving the DRAM particles by the PCIE device comprises: enabling the SRIOV function of the PCIE device; installing the physical function PF driver and the virtual function VF driver; implementing mapping of the PCIE address, the server address and the memory address, and writing the address mapping into the PF drive and VF drive.

Preferably, the method further includes: deploying the memory pool: controlling, by a management unit, a memory space of the shared memory pool of the plurality of servers; running the PF driver in the management unit, thereby driving the user space and the virtual function driving space The IDs correspond to and match; the virtual function driver is run on each server, so that the server finds its own corresponding address space and operates.

Preferably, the method further comprises: determining whether the server uses the allocated memory space, and if so, releasing the memory space.

Preferably, if it is determined that the requested memory space is not available, the method further includes: waiting, and determining whether there is a newly released memory space; if the released memory space satisfies the requested memory requirement, the released memory space is allocated. Give the server.

According to an embodiment of the present invention, an apparatus for dynamically allocating memory includes: a request receiving unit, configured to receive a memory allocation request to a server; and a determining unit, configured to perform, according to the memory allocation request, based on a memory pool composed of a plurality of memory elements driven by a standard device of the bus interface, determining whether the memory pool has a memory sum of one or more free memory particles satisfying the requested memory size; and an allocation unit for using the requested Memory is allocated to the server.

Preferably, the memory particles comprise DRAM particles; the device further comprises: an interface conversion unit for expanding the capacity of the DRAM interface through the memory buffer; and connecting the input of the memory controller to the DRAM particles, and performing DDR on the memory controller The conversion logic of the memory process to the PCIE process causes the output of the memory controller to be a PCIE interface, so that the output of the memory controller is a PCIE interface.

Preferably, the apparatus further includes: a driving unit configured to enable the SRIOV function of the PCIE device; installing the PF driver and the VF driver; and implementing mapping of the PCIE address, the server address, and the memory address, and writing the address mapping to the PF drive and VF drive.

Preferably, the device further includes: a memory pool deployment unit, configured to set a management unit to control a memory space of the plurality of server shared memory pools; and run the PF driver in the management unit to thereby user space and VF space The IDs correspond to and match; and the VF driver is run on each server, so that the server finds its own corresponding address space and operates.

Preferably, the determining unit is further configured to: determine whether the server uses the allocated memory space; and the device further includes: a releasing unit, configured to release the used memory space.

Preferably, the determining unit is further configured to: if it is determined that the requested memory space is not available, wait, and determine whether there is a newly released memory space; if the released memory space satisfies the requested memory requirement, indicate the The allocation unit allocates the freed memory space to the server.

According to an embodiment of an aspect of the present invention, a system for dynamically allocating memory is provided, the system comprising: a memory pool composed of a plurality of DRAM particles driven by a PCIE device; one or more servers; and, An apparatus for dynamically allocating memory.

In accordance with an embodiment of another aspect of the present invention, a memory is provided that includes a plurality of memory particles, wherein the memory particles are driven by a bus interface standard device.

It can be seen that the present invention separates the server and the memory through the PCIE through a memory pool composed of a plurality of DRAM particles driven by the PCIE device, and can realize dynamic allocation and on-demand allocation of memory by different servers through PCIE exchange. Preferably, the capacity expansion is performed by the memory buffer during the process of converting the interface of the DRAM particles into the PCIE interface. In addition, because memory granules are expanded by memory buffering, and through dynamic allocation and on-demand allocation of memory pools, there is no need to increase the entire memory stick compared to standard memory, so the cost has a lower advantage. Moreover, PCIE devices can be hot swapped compared to existing standard memory, and maintainability is enhanced.

Those skilled in the art will appreciate that although the following detailed description is made with reference to the illustrated embodiments and drawings, the invention is not limited to these embodiments. Rather, the scope of the invention is intended to be limited the scope of the invention

DRAWINGS

Other features, objects, and advantages of the present invention will become more apparent from the Detailed Description of Description

1 is a flow chart of a method of dynamically allocating memory according to an embodiment of the present invention;

2 is a schematic diagram of converting a set of DRAM interfaces into PCIE interfaces in a method for dynamically allocating memory according to an embodiment of the invention;

3 is a schematic diagram of a single-particle DRAM interface converted into a PCIE interface in a method for dynamically allocating memory according to an embodiment of the present invention;

4 is a schematic diagram of a DRAM pool deployment based on a PCIE interface in a method for dynamically allocating memory according to an embodiment of the present invention;

FIG. 5 is a schematic structural diagram of an apparatus for dynamically allocating memory according to an embodiment of the present invention.

detailed description

Before discussing the exemplary embodiments in more detail, it should be noted that some exemplary embodiments are described as a process or method depicted as a flowchart. Although the flowcharts describe various operations as a sequential process, many of the operations can be implemented in parallel, concurrently or concurrently. In addition, the order of operations can be rearranged. The process may be terminated when its operation is completed, but may also have additional steps not included in the figures. The processing may correspond to methods, functions, procedures, subroutines, subroutines, and the like.

The computer device includes a user device and a network device. The user equipment includes, but is not limited to, a computer, a smart phone, a PDA, etc.; the network device includes but is not limited to a single network server, a server group composed of multiple network servers, or a cloud computing based computer Or a cloud composed of a network server, wherein cloud computing is a type of distributed computing, a super virtual computer composed of a group of loosely coupled computers. Wherein, the computer device can be operated separately to implement the present invention, and can also access the network and implement the present invention by interacting with other computer devices in the network. The network in which the computer device is located includes, but is not limited to, the Internet, a wide area network, a metropolitan area network, a local area network, a VPN network, and the like.

It should be noted that the user equipment, the network equipment, the network, and the like are merely examples, and other existing or future possible computer equipment or networks, such as those applicable to the present invention, are also included in the scope of the present invention. It is included here by reference.

The methods discussed below, some of which are illustrated by flowcharts, can be implemented in hardware, software, firmware, middleware, microcode, hardware description languages, or any combination thereof. When implemented in software, firmware, middleware or microcode, the program code or code segments to carry out the necessary tasks can be stored in a machine or computer readable medium, such as a storage medium. The processor(s) can perform the necessary tasks.

The specific structural and functional details disclosed are merely representative and are for the purpose of describing exemplary embodiments of the invention. The present invention may, however, be embodied in many alternative forms and should not be construed as being limited only to the embodiments set forth herein.

It should be understood that although the terms "first," "second," etc. may be used herein to describe the various elements, these elements should not be limited by these terms. These terms are used only to distinguish one unit from another. For example, a first unit could be termed a second unit, and similarly a second unit could be termed a first unit, without departing from the scope of the exemplary embodiments. The term "and/or" used herein includes any and all combinations of one or more of the associated listed items.

It will be understood that when a unit is referred to as "connected" or "coupled" to another unit, it can be directly connected or coupled to the other unit, or an intermediate unit can be present. In contrast, when a unit is referred to as being "directly connected" or "directly coupled" to another unit, there is no intermediate unit. Other words used to describe the relationship between the units should be interpreted in a similar manner (eg "between" and "directly between" and "adjacent to" Than "directly adjacent to", etc.).

The terminology used herein is for the purpose of describing the particular embodiments, The singular forms "a", "an", It is also to be understood that the terms "comprising" and """ Other features, integers, steps, operations, units, components, and/or combinations thereof.

It should also be noted that, in some alternative implementations, the functions/acts noted may occur in a different order than that illustrated in the drawings. For example, two figures shown in succession may in fact be executed substantially concurrently or sometimes in the reverse order, depending on the function/acts involved.

The technical terms in the embodiments of the present invention are first explained as follows.

Memory granules: Also known as memory chips or memory chips, often referred to as memory cells.

DRAM granules: DRAM chips are used as memory granules.

A PCIE (Peripheral Component Interface Express) device refers to a device that supports the PCIE protocol.

The SRIOV feature allows efficient sharing of PCIE devices between devices/virtual machines.

VF (Virtual Function) driver is mainly used to discover PCIE devices.

The PF (Physical Function) driver is mainly used to manage the correspondence between the address of each user space in the memory and the ID of the VF.

The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings.

1 is a flow chart of a method of dynamically allocating memory in accordance with an embodiment of the present invention. The method of this embodiment mainly includes the following steps:

S110. Receive a memory allocation request of the server.

S120. Determine, according to a memory allocation request, whether the memory pool has one or more free memory particles to satisfy the requested memory space, based on a memory pool composed of a plurality of memory particles driven by the PCIE device.

If there is a requested memory space, step S130 is performed; if there is no requested memory space, step S140 is performed.

S130. Allocate the requested memory space to the server.

S140, waiting;

S150. Determine whether there is a newly released memory space, and determine whether the released memory space satisfies the requested memory requirement.

If the released memory space satisfies the requested memory requirement, execute S160; if not, return to S140 and continue to wait.

S160. Allocate the released memory space to the server.

The above memory pool is composed of a plurality of memory particles driven by PCIE devices, wherein the memory particles may be DRAM particles. This means that interface conversion and protocol conversion are required for DRAM particles, and the interface-converted DRAM particles need to be driven by PCIE devices, and the DRAM pool based on the PCIE interface needs to be deployed.

In order to further understand the present invention, the following aspects are further described in detail from the above aspects.

First, the implementation of the DRAM interface to the PCIE interface is introduced.

2 is a schematic diagram of converting a set of DRAM interfaces into PCIE interfaces in a method for dynamically allocating memory according to an embodiment of the invention.

In the specific implementation, the interface conversion and protocol conversion functions can be implemented by means of the FPGA.

The process of transferring a DRAM interface to a PCIE interface includes:

(Step 1) The DRAM interface is capacity-expanded by MEMORY BUFFER.

The memory mentioned in the present invention refers to server memory. Those skilled in the art understand that the server memory is also memory, and it has no obvious substantive difference between the appearance and the structure of the ordinary PC. It mainly introduces some new technologies into the memory.

For example, server memory can be divided into Buffer memory with cache and Unbuffer memory without cache. Buffer is a cache, which can also be understood as a cache. There are many applications in the server and graphics workstation memory, and the capacity is mostly 64K. However, as the memory capacity increases, its capacity also increases, and the memory with Buffer will be The read and write speed of the memory is greatly improved. Almost all memory with Buffer comes with ECC (Error Checking and Correcting).

Register is the register or directory register. The role of memory can be understood as the book directory. Through Register, when the memory is connected to the read and write instructions, this directory will be retrieved first, then read and write operations, which will greatly improve the server memory work. effectiveness. Memory with Register (register or directory register) must have a Buffer, and the Register memory that can be seen at present also has ECC function.

LRDIMMs (Load-Reduced DIMMs) reduce the server's memory bus load and power consumption by using new technologies and lower operating voltages, and allow the server memory bus to achieve higher operating frequencies and significantly increase Memory support capacity.

For a typical Unbuffered DIMM, the Registered DIMM used by the server boosts the memory support capacity by buffering the signal on the memory stick and relocating the memory granules. The LRDIMM memory changes the Register chip on the current RDIMM memory to an iMB (isolation Memory Buffer). The memory isolation buffer chip reduces the load on the memory bus and further increases the memory support capacity accordingly.

The DIMMs of the present invention are not limited in type, i.e., the DIMMs of the present invention cover current or future types of DIMMs. Also, the memory capacity is increased by the DIMM's Memory Buffer function.

After the above step (1), by connecting the high-speed IO pin of the FPGA to the DIMM interface and then defining these pins, the internal logic implementation simulates a MEMORY controller internally by the signals of these high-speed pins.

DRAM has the advantages of low power consumption, high integration (large single-chip capacity), and low price, but the control of DRAM is relatively complicated and requires timing refresh, so it is necessary to design a DRAM controller. The FPGA (Field Programmable Gate Array) provides large-capacity programmable logic to design a DRAM controller (MEMORY controller) that meets specific requirements. Since the FPGA is a CMOS process, its power consumption is very small, and at the same time, the FPGA can be rewritten to facilitate performance expansion. If necessary, simply change the internal logic of the FPGA to suit different design requirements or environmental requirements. In addition to using the FPGA to implement the memory controller, other programmable logic devices can be implemented, such as a CPLD (Complex Programmable Logic Device), a PLD (Programmable Logic Device), and the like.

(Step 2) The input of the MEMORY controller is connected to the DRAM pellet, so the input of the MEMORY controller is a DDR unit supporting a DDR (double data rate) process.

(Step 3) The exit of the MEMORY controller is a high-speed SERDES (SERializer/DESerializer) that supports NVME (Non-Volatile Memory Express) and PCIE (Peripheral Component Interface Express). The deserializer) matches the PCIE interface of the PCIE device.

In this example, DDR memory is taken as an example for description, which has the advantage of a fast transmission rate.

NVMe is a logical device interface standard like AHCI. It is a specification for SSDs using PCI-E channels. NVMe is designed to take full advantage of the low latency and parallelism of PCI-E SSD, as well as contemporary processing. Parallelism of devices, platforms and applications. The parallelism of SSD can be fully utilized by the hardware and software of the host. Compared with the current AHCI standard, the NVMe standard can bring various performance improvements.

PCIE is a high-speed serial point-to-point dual-channel high-bandwidth transmission. The connected devices allocate exclusive channel bandwidth and do not share bus bandwidth. They mainly support active power management, error reporting, end-to-end reliability transmission, hot swap, and quality of service ( QOS) and other functions. The main advantage of PCIE is the high data transfer rate. For example, the current highest 16X 2.0 version can reach 10GB/s, and there is considerable potential for development.

Here, the DDR, PCIE and NVME processes/protocols need to implement a complete set of logic for the logical translation of the hardware language. In this example, it is implemented inside the FPGA. In order to convert the DRAM interface to PCIE interface conversion, the FPGA needs to identify DDR, PCIE and NVME and perform logic conversion. Referring to FIG. 2, the FPGA internally includes a DDR unit, an NVMe unit, and a PCIE unit, wherein the DDR unit serves as an input, supports DDR Process, and is connected to a unified interface of a group of DRAM particles; the NVMe unit supports the NVM protocol, and connects the DDR unit and the PCIE unit. The PCIE unit supports the PCIE protocol, which acts as an FPGA output and provides a PCIE interface to the PCIe device.

In this way, a set of DRAM interfaces can be converted into a PCIE interface, for example, converted to a PCIE X8 interface.

Referring to FIG. 3, a schematic diagram of a single-particle DRAM interface converted into a PCIE interface in a method for dynamically allocating memory according to an embodiment of the present invention. The difference between the mode of Figure 3 and the mode of Figure 2 is that instead of changing the interface of a group of DRAM particles, the interface of a single DRAM particle is changed, both of which can achieve the same purpose.

The process of transferring a DRAM interface to a PCIE interface includes:

(Step 1) The DRAM interface is capacity-expanded by MEMORY BUFFER.

(Step 2) The input of the MEMORY controller is connected to the DRAM pellet, so the input of the MEMORY controller is a DDR unit supporting the DDR (double data rate) protocol.

(Step 3) The exit of the MEMORY controller is a PCIE unit supporting PCIE (Peripheral Component Interface Express) to match the PCIE interface of the PCIE device.

In a specific implementation, the above interface conversion can be implemented by means of a wafer. The conversion of the DRAM interface to the PCIE interface is completed in the DRAM chip package, that is, the interface of the DRAM secondary package is a PCIE interface chip.

Below, the implementation details of driving the converted DRAM particles using a PCIE device are described.

After the interface conversion shown in FIG. 2 or FIG. 3 above, the DRAM particles based on the PCIE interface have been obtained, and then the DRAM based on the PCIE interface needs to be driven, thereby sharing the memory capacity or the single server for multiple servers. Prepare for sharing the memory capacity of multiple virtual machines.

Those skilled in the art understand that SRIOV technology is a hardware-based virtualization solution that improves performance and scalability. The SRIOV standard allows for efficient sharing of PCIE devices between virtual machines, and it is implemented in hardware to achieve I/O performance comparable to native performance. The SRIOV specification defines a new standard by which new devices are created that allow virtual machines to be directly connected to I/O devices. A single I/O resource can be shared by many virtual machines. Shared devices will provide dedicated resources and also use shared common resources. This way, each virtual machine has access to a unique resource. Therefore, a PCIE device with SRIOV enabled and with appropriate hardware and OS support can be displayed as multiple separate physical devices, each with its own PCIE configuration space.

The two main functions in SRIOV are: (1) Physical Function (PF): PCI function to support SRIOV function, as defined in the SRIOV specification. The PF contains the SRIOV functional structure for managing SRIOV functions. PF is a full-featured PCIE feature that can be discovered, managed, and processed just like any other PCIE device. PF has fully configured resources that can be used to configure or control PCIE devices. (2) Virtual Function (VF): A function associated with a physical function. VF is a lightweight PCIE feature that shares one or more physical resources with physical functions and other VFs associated with the same physical function. VF only allows configuration resources for its own behavior.

In the present invention, driving the DRAM based on the PCIE interface mainly includes the following processes:

(1) First, enable the SRIOV function of the PCIE hardware.

This feature enable indicates that this PCIE device supports the SRIOV function and can be discovered at the OS level.

(2) PF drive

Installed on the server or a driver in the management machine. This driver is mainly used to manage the correspondence between the address of each user space in the memory and the ID of the VF. That is, the PF can see the address corresponding to the ID of all the space of the PCIE device. Manage.

(3) VF drive

A driver installed in a virtual machine, mainly used to discover PCIE devices.

(4) Address mapping

Mainly to achieve the mapping of PCIE address, server (virtual machine) address and memory address.

(5) Complete memory address mapping when loading the driver

These address maps are completed when the driver is loaded to see this PCIE space in memory.

Finally, the implementation details of the deployment of the DRAM pool based on the PCIE interface are introduced.

After completing the above interface conversion and PCIE interface-based DRAM driver, a memory pool composed of a plurality of DRAM particles based on the PCIE interface is obtained, and the memory pool needs to be properly deployed and managed to efficiently allocate memory. The purpose of the server sharing the same memory pool.

The deployment of the DRAM pool based on the PCIE interface mainly includes the following processes:

(1) Multiple physical machines share the memory space of the memory pool

Multiple hosts share the storage space in this memory pool.

(2) Management unit management PF driver operation

The PF driver is running in the management unit, which is to match the user space with the space ID of the VF and perform flexible online matching.

(3) Run VF driver on SERVER

After the above preparation is completed, the VF runs on each server, and the server finds the address space that it sees and can operate the space.

(4) The memory space allocation on each SERVER is managed by the management unit and is flexibly distributed online.

4 is a schematic diagram of a DRAM pool deployment based on a PCIE interface in a method for dynamically allocating memory according to an embodiment of the present invention.

In FIG. 4, a plurality of servers, a memory pool composed of a plurality of DRAM particles driven by PCI devices, a management unit, and a PCIE switch are shown. The server includes a PCIE module. As described above, the VF driver is run on the server, so that the server discovers its own address space and operates on the address space.

The management unit is responsible for managing the memory allocation of the server to the memory pool, including three aspects: managing the memory that is already in use; releasing the used memory; and not using the memory allocation. Specifically, the management unit allocates the memory capacity required for the request to the server according to the memory allocation request of the server, and after the usage is completed, releases and re-allocates to the server with the required request.

The PCIE switch provides multiple ports to connect multiple memory granules, so you can allocate memory space for multiple memory granules to a specific server at once.

Compared with the prior art, each server accesses a fixed amount of memory through a slot, not only expands the memory capacity through the memory buffer, but also realizes memory allocation as needed. For example, a server has 16 slots. If the capacity of each memory module is 16G, even if it is fully populated, it only has 256G capacity. If it is required to have 300G memory, only one server is added according to the existing method, although this can be satisfied. Memory requirements, however new server additions Other resources are wasted. In the manner of the present invention, the memory capacity is expanded by the memory buffer, and a certain number of DRAM particles driven by the PCIE device can be set according to requirements. When the server makes a memory allocation request, the memory can be dynamically selected according to the requested memory size. The memory granules requesting the amount of memory size, the content of the part of the memory granules is allocated to the server, and after the server is used, the memory space is dynamically released for use by other servers in need.

It can be seen that the present invention converts the DRAM interface to the PCIE interface by converting the interface of the DRAM particles into a PCIE interface through a programmable logic chip (FPGA) or a chip, and then designating the allocated PCIE device through the driver of the PCIE device. After a specific compute node, the PCIE address of this part is mapped to the memory address, thereby realizing the purpose of driving the allocated portion of the memory; through the deployment and management of the memory pool, the server can be allocated a certain number of requests. Memory space, and dynamic management of memory allocation and release.

Compared with the prior art, the solution of the invention has at least the following advantages:

(1) Allocate memory on demand

Through the memory pool composed of DRAM particles driven by PCIE devices, the dynamic allocation and on-demand allocation of memory by different servers can be realized through PCIE exchange.

(2) Separation of memory and server

PCIE expansion can be achieved through the conversion of the PCIE interface. The PCIE is connected to the PCIE SLOT of the standard server, so that the server and the memory are separated by PCIE.

(3) achieve large memory

With the expansion of the memory BUFFER, the capacity can be expanded many times compared to the standard memory.

(4) Cost savings

Since the memory granules are expanded by the memory buffer and the dynamic allocation and on-demand allocation of the memory pool, the cost is lower than that of the standard memory.

(5) Maintainability enhancement

PCIE devices can be hot swapped compared to existing standard memory, and maintainability is enhanced.

An embodiment of the present invention provides an apparatus for dynamically allocating memory corresponding to the foregoing method. Referring to Figure 5, the device includes:

The request receiving unit 501 is configured to receive a memory allocation request of the server;

The determining unit 502 is configured to determine, according to the memory allocation request, based on a memory pool composed of a plurality of memory particles driven by the PCIE device, whether the memory pool has one or more free memory particles and the memory sum meets the request The size of the memory;

The allocating unit 503 is configured to allocate the requested memory to the server.

Preferably, the device further comprises:

The interface conversion unit 504 is configured to expand the capacity of the DRAM interface through the memory buffer; and connect the input of the memory controller to the DRAM particles, and perform the conversion logic of the DDR memory process to the PCIE process in the memory controller to make the output of the memory controller For the PCIE interface, the output of the memory controller is the PCIE interface.

Preferably, the device further comprises:

The driving unit 505 is configured to enable the SRIOV function of the PCIE device; install the PF driver and the VF driver; and implement mapping of the PCIE address, the server address and the memory address, and write the address mapping to the PF driver and the VF driver.

Preferably, the device further comprises:

The memory pool deployment unit 506 is configured to set a management unit to control a memory space of the shared memory pool of the plurality of servers; and run the PF driver in the management unit to match and match the user space with the VF space ID; and The VF driver is run on each server, so that the server finds its own address space and operates.

Preferably,

The determining unit 502 is further configured to: determine whether the server uses the allocated memory space;

The device further includes a release unit 507 for releasing the used memory space.

Preferably,

The determining unit 502 is further configured to: if it is determined that the requested memory space is not available, wait, and determine whether there is a newly released memory space; if the released memory space satisfies the requested memory requirement, the indicating allocation unit 503 is released. The memory space is allocated to the server.

In addition, the present invention also provides a system for dynamically allocating memory, the system comprising: a memory pool composed of a plurality of memory particles driven by PCIE devices; one or more servers; and, as shown in FIG. 5 described above A device that dynamically allocates memory.

In addition, the present invention also provides a memory including a plurality of memory particles, wherein the memory particles are driven by a PCIE device.

It should be noted that the present invention can be implemented in software and/or a combination of software and hardware, for example, using an application specific integrated circuit (ASIC), a general purpose computer, or any other similar hardware device. In one embodiment, the software program of the present invention may be executed by a processor to implement the steps or functions described above. Similarly, this The inventive software program (including related data structures) can be stored in a computer readable recording medium such as a RAM memory, a magnetic or optical drive or a floppy disk and the like. Additionally, some of the steps or functions of the present invention may be implemented in hardware, for example, as a circuit that cooperates with a processor to perform various steps or functions.

Additionally, a portion of the invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide a method and/or solution in accordance with the present invention. The program instructions for invoking the method of the present invention may be stored in a fixed or removable recording medium and/or transmitted by a data stream in a broadcast or other signal bearing medium, and/or stored in a The working memory of the computer device in which the program instructions are run. Herein, an embodiment in accordance with the present invention includes a device including a memory for storing computer program instructions and a processor for executing program instructions, wherein when the computer program instructions are executed by the processor, triggering The apparatus operates based on the aforementioned methods and/or technical solutions in accordance with various embodiments of the present invention.

It is apparent to those skilled in the art that the present invention is not limited to the details of the above-described exemplary embodiments, and the present invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the present embodiments are to be considered as illustrative and not restrictive, and the scope of the invention is defined by the appended claims instead All changes in the meaning and scope of equivalent elements are included in the present invention. Any reference signs in the claims should not be construed as limiting the claim. In addition, it is to be understood that the word "comprising" does not exclude other elements or steps. A plurality of units or devices recited in the system claims can also be implemented by a unit or device by software or hardware. The first, second, etc. words are used to denote names and do not denote any particular order.

Claims

A method for dynamically allocating memory, comprising:

Receiving a memory allocation request from the server;

Determining, according to the memory allocation request, based on a memory pool composed of a plurality of storage particles driven by the bus interface standard device, determining whether the memory pool has a memory sum of one or more free storage particles satisfying the requested memory size;

If so, the requested memory is allocated to the server.
The method of claim 1 wherein said storage particles comprise DRAM particles, said method further comprising: converting the DRAM interface of the DRAM particles into a PCIE interface.
The method of claim 2 wherein said converting the DRAM interface of the DRAM particles to a PCIE interface comprises:

Capacity expansion of the DRAM interface through memory buffering;

The input of the memory controller is connected to the DRAM particles, and the memory controller performs a conversion logic of the double-rate DDR memory process to the PCIE process, so that the output of the memory controller is a PCIE interface.
The method of claim 1 wherein driving the DRAM particles by the PCIE device comprises:

Enable the SRIOV function of the PCIE device.

Install physical function PF driver and virtual function VF driver;

A mapping of the PCIE address, the server address, and the memory address is implemented, and the address mapping is written to the PF driver and the VF driver.
The method of claim 4, further comprising: deploying the memory pool:

Controlling the memory space of the shared memory pool of multiple servers by a management unit;

Running the PF driver in the management unit, thereby correspondingly matching and matching the user space with the virtual function driving space ID;

The virtual function driver is run on each server, so that the server finds its own corresponding address space and operates.
The method of any of claims 1 to 5, further comprising:

Determine whether the server uses the allocated memory space, and if so, release the memory space.
The method of claim 6, wherein if it is determined that the requested memory space is not available, the method further comprises:

Wait and determine if there is a newly released memory space;

If the freed memory space meets the requested memory requirement, the freed memory space is allocated to the server.
A device for dynamically allocating memory, comprising:

a request receiving unit, configured to receive a memory allocation request of the server;

a determining unit, configured to determine, according to the memory allocation request, whether the memory pool has a memory sum of one or more free memory particles based on a memory pool composed of a plurality of memory particles driven by a bus interface standard device The requested memory size;

An allocation unit for allocating the requested memory to the server.
The device of claim 8 wherein said memory particles comprise DRAM particles; said apparatus further comprising:

The interface conversion unit is configured to expand the capacity of the DRAM interface through the memory buffer; and, the input of the memory controller is connected to the DRAM particle, and the conversion logic of the DDR memory process to the PCIE process is performed in the memory controller, so that the output of the memory controller is The PCIE interface makes the output of the memory controller a PCIE interface.
The device of claim 8 further comprising:

The driving unit is configured to enable the SRIOV function of the PCIE device; install the PF driver and the VF driver; and implement mapping of the PCIE address, the server address and the memory address, and write the address mapping to the PF driver and the VF driver.
The device of claim 10, further comprising:

a memory pool deployment unit, configured to set a management unit to control a memory space of the shared memory pool of the plurality of servers; and run the PF driver in the management unit to match and match the user space with the VF space ID; and The VF driver is run on each server, so that the server finds its own corresponding address space and operates.
A device according to any one of claims 8 to 10, characterized in that

The determining unit is further configured to: determine whether the server uses the allocated memory space;

The device further includes a release unit for releasing the used memory space.
The device of claim 12 wherein:

The determining unit is further configured to: if it is determined that the requested memory space is not available, wait, and determine whether there is a newly released memory space; if the released memory space satisfies the requested memory requirement, indicate that the allocating unit The freed memory space is allocated to the server.
A system for dynamically allocating memory, comprising: driving by a plurality of standard devices via a bus interface A memory pool of memory particles; one or more servers; and the apparatus for dynamically allocating memory according to any one of claims 8-13.
A memory, characterized in that the memory comprises a plurality of memory particles, wherein the memory particles are driven by a bus interface standard device.