CN103501242B - Method for managing resource and multiple-node cluster device - Google Patents
Method for managing resource and multiple-node cluster device Download PDFInfo
- Publication number
- CN103501242B CN103501242B CN201310430685.9A CN201310430685A CN103501242B CN 103501242 B CN103501242 B CN 103501242B CN 201310430685 A CN201310430685 A CN 201310430685A CN 103501242 B CN103501242 B CN 103501242B
- Authority
- CN
- China
- Prior art keywords
- node
- host
- destination
- service managerz
- business
- 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
Landscapes
- Multi Processors (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The present invention relates to a kind of method for managing resource and multiple-node cluster device.The method includes:The service managerZ-HU of multiple-node cluster device determines the type of node;If the node is the host node, the virtual substrate controller of the node is set up, according to resource allocation strategy, configure the membership and routing iinformation of the host node;If the node is the destination node, according to resource allocation strategy, the membership of the destination node is configured;Electricity on the business main power source of the node is controlled, so that the node can run business.Node can be managed collectively by virtual substrate controller, the wasting of resources can be reduced according to different resource allocation strategy, flexible configuration resource.
Description
Technical field
The present invention relates to communication technical field, more particularly to a kind of method for managing resource and multiple-node cluster device.
Background technology
Due to there are a large amount of different workloads in data center(workload)Scene, under different application scenarios
The problem that deployment may be many is carried out using identical server.Firstly, it is necessary to reserved more resource(As reserved compared with multiple disks
And I/O interface card space), to meet the needs of different users, but in actual deployment possibly also with less than.Secondly, needed in user
During a certain item vast resources, it has to pay more extra costs, such as routine 2P/2U Xeon to the big multi-configuration of strong server
24 disks, when user needs 30 disk sizes, user needs one 2P server of extra purchase.Additionally, also wasting a large amount of
Energy consumption, such as user buy 2 servers, it has to complete power supply is all disposed for two-server to meet 30 disk demands
Consumption.
In practical application, different servers can be configured according to different application scenarios.For compute-intensive applications, can
To configure high-performance processor and a small amount of disk;For storage intensive application, more disks can be configured, it is bigger to provide
Memory capacity.FBCI(Fabric Based Computing Infrastructure, UNE framework)Can be based on
PCIe(Peripheral Component Interconnect Express, the interconnection of high-speed peripheral component)、IB
(infiniband)、FCoE(Fibre Channel over Ethernet, Ethernet optical-fibre channel)Deng UNE technology
To strong server(Xeon)Framework.Combined with virtual technology, can be by NIC by UNE(Network Interface
Card, network adapter)With disk(Disk)Multiple VM is given in resource-sharing(Virtual Machine, virtual machine), reach lifting
The purpose of resource utilization.Such as Huawei E9000FusionCube and IBM PureFlex.But, FBCI cannot directly share
Disk resource is, it is necessary to extra protocol conversion process unit such as HBA(Host Bus Adapter, host bus adaptor)Ability
Exchange network, increased system cost and complexity.Also, because the granularity of node is larger, needing the number of nodes of deployment
It is larger as more than 50 when, it is necessary to UNE in exchange(Fabric)Network size is larger, and cost is also increased.
In order to reduce node granularity, can be by multiple resources such as disk, internal memory(Memory)、IO(Input-Output, it is defeated
Enter output)Etc. the node for being integrated into.Such as 1 node includes 1 processor and 1 disk(1P+1Disk), or 1
Individual node includes 1 processor and 2 disks(1P+2DISK).Interconnected by Ethernet and constitute microserver node cluster
(Cluster).
However, such a granularity still cannot meet demand of the user to small grain size.
The content of the invention
Technical problem
In view of this, soluble technical problem of the invention is how to reduce resource granularity.
Solution
An embodiment of the invention, first aspect, there is provided a kind of method for managing resource, are applied to and data center
Operation system connection multiple-node cluster device in, the node of the multiple-node cluster device includes host node Host and target
Node Target, the host node includes processor and internal memory;The destination node includes disk, input-output equipment, network
At least one or various in adapter and acceleration component, methods described includes:
The service managerZ-HU of the multiple-node cluster device determines the type of the node;
If the node is the host node, the service managerZ-HU sets up the virtual substrate controller of the node,
According to resource allocation strategy, the membership and routing iinformation of the host node are configured, the virtual substrate controller is used to manage
Manage the host node and there is the destination node of membership with the host node, the routing iinformation arrives it for the host node
Business route between his host node;
If the node is the destination node, the service managerZ-HU configures the mesh according to resource allocation strategy
Mark the membership of node;
The service managerZ-HU controls electricity on the business main power source of the node, so that the node can run business.
With reference in a first aspect, in the first possible implementation, in the mistake of multiple-node cluster device initialization
In journey or when there is the node of insertion in need, before the service managerZ-HU determines the node type of the node, bag
Include:
It is electricity, the auxiliary tube on the accessory power supply of the Added Management unit of all nodes of the multiple-node cluster device
Reason unit includes accessory power supply, memory cell and interface unit, wherein, preserved in the memory cell for identifying the section
Point is host node or the electronic labeling information of destination node, and the interface unit is connected with the service managerZ-HU;
The service managerZ-HU determines the type of the node, including:
The service managerZ-HU reads the electronic labeling information from the memory cell of the Added Management unit, according to
The electronic labeling information determines the type of the node.
With reference to the first possible implementation of first aspect, in second possible implementation, described many
During node cluster device is initialized, the service managerZ-HU controls electricity on the business main power source of the node, including:
The service managerZ-HU controls electricity on the business main power source of the destination node;
After electricity on the business main power source of the destination node, the service managerZ-HU controls the business of the host node
Electricity on main power source.
With reference to first aspect or its any one possible implementation, in the third possible implementation, the clothes
Business manager is set up after the virtual substrate controller of the node, including:
Whether the virtual substrate controller checks the host node normal, and according to the resource allocation strategy, inspection
Have the destination node of membership whether normal with the host node.
With reference to first aspect or its any one possible implementation, in the 4th kind of possible implementation, described
After node start to process business, if desired by the knot removal, methods described also includes:
The operation system of the data center discharges the business of the node;
The service managerZ-HU determines type and groove the position information of the node;
If the node is the host node, the service managerZ-HU discharges the virtual substrate controller of the node,
And record deletion daily record, by electricity under the node;
If the node is the destination node, by electricity under the node;
The node according to the information removing of groove position.
With reference to first aspect or its any one possible implementation, in the 5th kind of possible implementation, the master
Node and the destination node insert the backboard of the multiple-node cluster device by unified interface connector;
The multiple-node cluster device also includes I/O interface layer and operation exchange network, and the operation exchange network passes through institute
Backboard is stated to be connected respectively with I/O interface layer, the host node and the destination node;
The service managerZ-HU is by the backboard and I/O interface layer, the operation exchange network, the host node
Connected respectively with the destination node.
A kind of second aspect, there is provided multiple-node cluster device, including:
I/O interface layer, for being connected with external equipment;
Operation exchange network, is connected respectively with I/O interface layer and mixing resource pool;
The mixing resource pool includes multiple nodes, and the node includes host node and destination node, the host node bag
Include processor and internal memory;The destination node include disk, input-output equipment, network adapter and acceleration component at least
One or more;
Service managerZ-HU, is connected respectively with I/O interface layer, the operation exchange network, the mixing resource pool, uses
In it is determined that the type of the node;If the node is the host node, the virtual substrate controller of the node, root are set up
According to resource allocation strategy, the membership and routing iinformation of the host node are configured, the virtual substrate controller is used to manage
The host node and there is the destination node of membership with the host node, the routing iinformation arrives other for the host node
Business route between host node;If the node is the destination node, according to resource allocation strategy, the target is configured
The membership of node;Electricity on the business main power source of the node is controlled, so that the node can run business.
With reference to second aspect, in the first possible implementation, the multiple-node cluster device also includes:
Power supply, divides with I/O interface layer, the operation exchange network, resource pool, the service managerZ-HU of mixing
Do not connect, for being initialized in the multiple-node cluster device during or when there is the node of insertion in need,
It is electricity on the Added Management unit of all nodes before the service managerZ-HU determines the node type of the node, it is described auxiliary
Helping administrative unit includes accessory power supply, memory cell and interface unit, wherein, preserved for identifying in the memory cell
It is host node or the electronic labeling information of destination node to state node, and the interface unit is connected with the service managerZ-HU;
The service managerZ-HU is additionally operable to read the electronic tag letter from the memory cell of the Added Management unit
Breath, the type of the node is determined according to the electronic labeling information.
With reference to the first possible implementation of second aspect, in second possible implementation, the service
Manager, is additionally operable to during the multiple-node cluster device is initialized, control the business main power source of the destination node
Upper electricity;After electricity on the business main power source of the destination node, electricity on the business main power source of the host node is controlled.
With reference to second aspect or its any one possible implementation, in the third possible implementation, the master
Node includes:
Whether the virtual substrate controller is normal for checking the host node;Inspection has with the host node is subordinate to
Whether the destination node of category relation is normal.
With reference to second aspect or its any one possible implementation, in the 4th kind of possible implementation, the clothes
Business manager, is additionally operable to after the node start to process business, if desired by the knot removal, in notifying the data
The operation system of the heart discharges the business of the node;Determine type and groove the position information of the node;If the node is described
Host node, then the virtual substrate controller of the service managerZ-HU release node, and record deletion daily record, by the node
Lower electricity;If the node is the destination node, by electricity under the node;The node according to the information removing of groove position.
With reference to second aspect or its any one possible implementation, in the 5th kind of possible implementation, also include
Backboard;
The operation exchange network is connected respectively by the backboard with I/O interface layer, the mixing resource pool;
The service managerZ-HU by the backboard and I/O interface layer, the operation exchange network, described mix money
Source pond connects respectively.
With reference to second aspect or its any one possible implementation, in the third possible implementation, the section
Point has 3.5 inch disk sizes, and the node accesses backboard by unified interface connector.
Beneficial effect
The embodiment of the present invention, multiple-node cluster device can reduce section using host node and destination node as management granularity
Point granularity.
According to below with reference to the accompanying drawings to detailed description of illustrative embodiments, further feature of the invention and aspect will become
It is clear.
Brief description of the drawings
Comprising in the description and the part that constitutes specification accompanying drawing together illustrated with specification it is of the invention
Exemplary embodiment, feature and aspect, and for explaining principle of the invention.
Fig. 1 a are the flow chart of the method for managing resource of the embodiment of the present invention one;
Fig. 1 b mix the schematic diagram of resource for the disc type in the method for managing resource of the embodiment of the present invention one;
Fig. 1 c are the schematic diagram of operation exchange network in the method for managing resource of the embodiment of the present invention one;
Fig. 2 is the flow chart of the method for managing resource of the embodiment of the present invention two;
Fig. 3 is the flow chart of the method for managing resource of the embodiment of the present invention three;
Fig. 4 is the flow chart of the method for managing resource of the embodiment of the present invention four;
Fig. 5 is the structured flowchart of the multiple-node cluster device of the embodiment of the present invention five;
Fig. 6 is the structured flowchart of the multiple-node cluster device of the embodiment of the present invention six.
Specific embodiment
Various exemplary embodiments of the invention, feature and aspect are described in detail below with reference to accompanying drawing.It is identical in accompanying drawing
Reference represent the same or analogous element of function.Although the various aspects of embodiment are shown in the drawings, remove
Non-specifically is pointed out, it is not necessary to accompanying drawing drawn to scale.
Special word " exemplary " means " being used as example, embodiment or illustrative " herein.Here as " exemplary "
Illustrated any embodiment should not necessarily be construed as preferred or advantageous over other embodiments.
In addition, in order to better illustrate the present invention, numerous details are given in specific embodiment below.
It will be appreciated by those skilled in the art that without some details, the present invention can equally be implemented.In some instances, for
Method well known to those skilled in the art, means, element and circuit are not described in detail, in order to highlight purport of the invention.
In the embodiment of the present invention, host node(Host)It can be the node with computing capability(Section can referred to as be calculated
Point), can typically include processor(CPU)And internal memory(Memory).Destination node(Target)Can include that there is storage energy
The node of power(Can abbreviation memory node)Such as disk(Disk), or including acceleration component such as FPGA(Field-
Programmable Gate Array, field programmable gate array)、GPU(Graphic Processing Unit, at figure
Reason device)、DSP(Digital singnal processor, digital signal processor)Deng input and output can also be included(IO)
Equipment, network adapter(NIC)Deng.From the point of view of another angle, host node can be the device for running group service OS;Mesh
Mark node can be the I/O equipment for not running group service OS, and destination node is shown as " equipment " in interface.Another angle
From the point of view of degree, host node has BMC(Basic Motherboard Controller, baseboard controller), and destination node is useless
BMC。
Compared to the prior art, embodiment of the present invention interior joint granularity is smaller.In other words, 1 node of the prior art
At least 2 nodes in the embodiment of the present invention can be split into.For example include 1 processor and 2 disks in the prior art
Node(1P+2Disk)In embodiments of the present invention, at least 2 nodes can be split into.Wherein 1 is host node, including 1
Processor(And the additional device such as internal memory, bus, interface);2 disks can constitute 1 destination node, this 1 target section
Point has 2 disks, or splits into 2 destination nodes, and each destination node possesses a disk.If for example, 2 disks are all
It is 2.5 cun of disks, then collectively constitutes 1 destination node, if 2 disks are all 3.5 cun of disks, splits into 2 destination nodes.
In the prior art, the slot needed to be inserted on backboard when host node and destination node run, host node and target
The slot of destination node between host node between node, different, different on backboard is often also different.
Insert the slot on backboard respectively due to host node and destination node, no longer integrate in the same node, they
Become relatively independent.Because host node has BMC, and destination node does not have BMC, causes host node and destination node in way to manage
Upper formation difference.For example, for host node, management software needs to collect the information of each BMC, but destination node need not be held
This is operated row, when the number of nodes in whole storage system is larger, such as when looking into 30 host nodes, and the collection of this information
Obvious influence can be caused to whole system resource.
The embodiment of the present invention eliminates the BMC of host node, and host node is without BMC in also just saying the embodiment of the present invention.This
For system administration brings convenience, system resource consuming is reduced.
Wherein, it is not provided with BMC in host node(Basic Motherboard Controller, baseboard controller), so
Avoid the need for setting the interface matched with BMC for host node, in section Example of the present invention, host node and destination node can lead to
Cross the backboard that unified interface connector inserts multiple-node cluster device.Wherein, unified interface connector can be size on backboard
With function identical groove position, the support insertion of this groove position different host nodes and different destination nodes can be by detections
The type of the node being inserted into is the corresponding resource of node distribution starts the node device being inserted into.Wherein, can be with backboard
With various grooves position, host node and destination node can share identical groove position, and this shared groove position is referred to as unified interface and connects
Device is connect, additionally, operation exchange network, service managerZ-HU, basic function component etc. also have corresponding groove position respectively in back.
In the prior art, in order to meet 30 demands of disk of client, user can load flexible configuration according to real work
15 1P+2Disk nodes, or 30 1P+1Disk nodes.But, the proportioning of storage resource and computing resource in this node
It is fixed, in allocating system resource, constrained by node resource proportioning.A kind of optional mode provided in an embodiment of the present invention
In, host node and destination node increased the flexibility of system using the groove position of same type, and each groove position can insert any one
Node is planted, can flexibly be matched.
Embodiment 1
Fig. 1 a are the flow chart of the method for managing resource of the embodiment of the present invention one.As shown in Figure 1a, the method for managing resource
In can apply to the multiple-node cluster device being connected with the operation system of data center, the node of the multiple-node cluster device
Host node and destination node can be included, the host node includes processor and internal memory;The destination node includes disk, input
At least one or various in output equipment, network adapter and acceleration component, methods described can specifically include following step
Suddenly.
Step 101, the service managerZ-HU of multiple-node cluster device determine the type of the node, if the node is described
Host node(Host), then step 102 is performed, if the node is the destination node(Target), then step 103 is performed.
In multiple-node cluster device, node type includes host node(Host)And destination node(Target).The main section
Point is made up of processor and internal memory, can also be including the other assemblies such as bus, external interface;The destination node can be storage
Equipment(Such as disk), it is input-output equipment, at least one or various in network adapter and acceleration component, for example:One
Individual destination node can be single disk, or multiple disk combination together, or an input-output equipment with
One network adapter is combined.
Memory cell can be set on host node and service node, preserved in the memory unit for identifying the section
Point is host node or the electronic labeling information of destination node.Service managerZ-HU can be determined that by reading this electronic labeling information
Egress is host node or destination node.It is the power supply of 3.3V that such as voltage can be provided to node, it is ensured that node energy
Enough by SP(Service Professor, service managerZ-HU)Identification.
Step 102, service managerZ-HU set up the virtual substrate controller of the node(Virtral BMC, vBMC).
Wherein, the virtual substrate controller is used to manage the host node and have membership with the host node
Destination node.Specifically, the virtual substrate controller can have the target of membership to host node and with host node
Node carries out membership configuration, power management, hardware configuration, node failure management etc..
Step 103, service managerZ-HU complete configuration according to resource allocation strategy.
Wherein, service managerZ-HU can configure the membership and routing iinformation of the host node, and the routing iinformation is
The host node route to the business between other host nodes.Additionally, service managerZ-HU can be configured according to resource allocation strategy
The membership of the destination node.Wherein, resource allocation strategy can pre-set according to user's request, for example:Target section
Point 2~10 belongs to host node 1, wherein, destination node 2~9 is disk, and destination node 10 is acceleration component;Host node 1, main section
Point 11, host node 19 can carry out operation exchange, it is necessary to set business route.
Step 104, service managerZ-HU control electricity on the business main power source of the node, so that the node can run industry
Business.
Wherein, the sequential relationship of step 102 and step 103 is not limited, it is also possible to is first carried out step 103 and is performed step again
102, or the two execution simultaneously.Additionally, for some destination nodes, it is also possible to set up virtual substrate controller.
Alternatively, embodiment of the present invention interior joint can be 3.5 "(I.e. 3.5 inches)Disk, 2.5 " disk, network interface card, PCIe
Equipment etc., the granularity of such node is small, belongs to a kind of microserver, low in energy consumption.Multiple host nodes and destination node can be constituted
Disc type mixes resource.Consider node density utilization rate, the size of node can be unified, and each node can be by system
One interface connector inserts the backboard of service managerZ-HU;Additionally, multiple-node cluster device can also include I/O interface layer and business
Exchange network, the operation exchange network is by the backboard and I/O interface layer, the host node and the destination node
Connect respectively;The service managerZ-HU is by the backboard and I/O interface layer, the operation exchange network, the host node
Connected respectively with the destination node.For example:Fig. 1 b are the disc type mixing money in the method for managing resource of the embodiment of the present invention one
The schematic diagram in source, as shown in Figure 1 b, the size of node shell can be used uniformly across about one 3.5 " disk size is (such as:About 147
× 101 × 26mm, is chosen as positive and negative deviation 5mm or so).In implementing, intra-node can include 1 3.5 " disk 21,
Or 2 2.5 " disk 22 or 1 calculate node 23 or 1 is including disk and etc. the mainboard of the node with computing capability 24,
Intra-node can also be including NIC or IO etc., and various types of nodes can be connected to backboard by unified interface connector
On.Wherein, in order to directly use standard disk, for the memory node for only including storage capacity(Such as disk 21,22)If, magnetic
The interface connector 26 of disk uses existing standard interface, then can also be by adaptation module(Adapter)25 are connected to disk
Unified interface connector 27, it is ensured that 3.5 " disk or other memory nodes being capable of common groove positions.Wherein, calculate node 23 can be wrapped
Include n(n=1,2,3…)Group processor and internal memory, node 24, can also be empty according to node layout except including processor and internal memory
Between flexible configuration small size storage medium such as mSATA1.8 " or 2.5 " disk.
As illustrated in figure 1 c, be the embodiment of the present invention one method for managing resource in operation exchange network schematic diagram, wherein,
Host node(Host)To run the root device of business software, such as calculate node;Destination node(Target)It is disk, IO, NIC
Or acceleration component etc..Assuming that the groove position of the total n+m node of multiple-node cluster device, during actual deployment, according to workload
(workload)Demand is, it is necessary to dispose n host node, and m destination node, wherein n, m=0,1,2,3,4 ....
Wherein, host node can share the groove position of multiple-node cluster device with destination node, and SP can provide n+m unification
Management interface, for connecting each node by backboard.SP can also provide operation exchange network(Fabric)In it is right
The multiple interfaces answered.Wherein, the system administration network of SP can be used but be not limited to SMbus(System Management
Bus, System Management Bus), RS485 standards or self-defined network type, can be hung in each management interface cabling branch
Multiple Node Slots are connect, to save system port quantity, simplifies Interface design.
It should be noted that because any one groove position of multiple-node cluster device there may be various non intelligent nodes
Such as disk, I/O device, also unified on the primary node no longer to provide independent in order to the resource to each node is managed collectively
BMC, and the SP on backboard, management multiple vBMC(Visual BMC, virtual BMC), vBMC is soft in the management system of SP
Part functional module.Wherein, a vBMC can correspond to a host node(Host), n host node be accomplished by n vBMC.SP pins
To each host node(Host)Independent vBMC admin instances are set up, equipment control and the information of node is realized.So as to unite
One management, while also reducing the management cost of host node.So, host node inside includes CPU, typically CPU+ internal memories
(Memory)Structure, operating system can be run(Operating System, referred to as:OS), BMC is not provided with host node,
But managed by vBMC.Destination node(Target)Such as disk(Disk)Or network interface card(NIC), there is no BMC, it is impossible to operation behaviour
OS, the vBMC management of the host node belonged to by it.Certainly, if application scenarios need, it is also possible to in some destination nodes
The intelligent node being such as made up of acceleration component, sets up independent vBMC in SP.
The method for managing resource of the present embodiment, multiple-node cluster device passes through virtual substrate controller(vBMC)Can be to section
Point is managed collectively, can be according to different resource allocation strategies, flexible configuration resource, reduces the wasting of resources.Additionally, target
The resources such as IO, storage, acceleration, NIC in node can share to all host nodes by operation exchange network.Also, divide
Different types of node(Such as calculate node or memory node), node granularity can be reduced, it is further possible to realize depositing entirely
The density of storage node or full calculate node maximizes configuration.
Embodiment 2
Fig. 2 is the flow chart of the method for managing resource of the embodiment of the present invention two, as shown in Fig. 2 the area with above-described embodiment
It is not that, in the power-up initializing stage of multiple-node cluster device, the method for managing resource can include:
Step 201, be in the multiple-node cluster device electricity.In the power on or power switch of multiple-node cluster device
After opening, electricity in multiple-node cluster device(Power UP).
Step 202, be on the service managerZ-HU electricity.Wherein, service managerZ-HU(SP)Electricity, operational management can preferentially be gone up
Service.
Step 203, be on the Added Management unit of all nodes of the multiple-node cluster device electricity.Wherein, it is described auxiliary
Helping administrative unit can include accessory power supply, memory cell and interface unit, wherein, preserved for marking in the memory cell
It is host node or the electronic labeling information of destination node to know the node, and the interface unit is connected with the service managerZ-HU.
The accessory power supply of the Added Management unit of all nodes generally only has 3.3V, and power consumption is very small, it is ensured that node can be by SP
Identification, but node is also not switched on business main power source, so business can't normally be run.
Step 204, the service managerZ-HU of multiple-node cluster device determine the type of the node, if the node is described
Host node, then perform step 205, if the node is the destination node, performs step 206.Specifically, service managerZ-HU
The electronic labeling information can be read from the memory cell of the Added Management unit of node, according to the electronic labeling information
Determine the type of the node.Additionally, service managerZ-HU can also obtain the corresponding groove position information of node, to facilitate follow-up determination
The actual corresponding physical slot position of node.
Step 205, service managerZ-HU set up the virtual substrate controller of the node.Checked by virtual substrate controller and saved
Whether point is normal.
Wherein, whether the virtual substrate controller of host node can check the host node normal, can also be according to described
Whether resource allocation strategy inspection has the destination node of membership normal with the host node.
Step 206, service managerZ-HU complete configuration according to resource allocation strategy.Wherein, service managerZ-HU can configure institute
The membership and routing iinformation of host node are stated, the routing iinformation is the host node to the business road between other host nodes
By.Additionally, service managerZ-HU according to resource allocation strategy, can configure the membership of the destination node.Wherein, if adopted
With with business interface such as SAS(Serial Attached Small Computer System Interface, serial small
Computer system interface)、SATA(Serial Advanced Technology Attachment, Serial Advanced Technology Attachment)
The backboard of the insertion multiple-node cluster device such as Switching Module of interface and PCIe interface forms operation exchange network.For specific
Interface type such as PCIe interface, in addition it is also necessary to configure the port type information of each node.
Wherein, the sequential relationship between step 205 and step 206 is not limited, it is also possible to is first carried out step 206 and is performed again
Step 205, or the two execution simultaneously.
After step 205 or step 206, the method for managing resource can also include:
Step 207, service managerZ-HU control electricity on the business main power source of the destination node.
Step 208, service managerZ-HU control electricity on the business main power source of the host node.
Wherein it is possible to be first electricity on the business main power source of destination node, the electricity on the business main power source of the destination node
Afterwards, then be on the business main power source of host node electricity.So, the virtual substrate controller of host node just can enumerating in host node
Stage, completion belongs to the hot plug of the destination node of the host node, can improve the speed of multiple-node cluster device initialization.
It is of course also possible to first carry out step 208 performs step 207 again, multiple-node cluster device also being capable of normal initialization.
After step 209, upper electric current journey terminate, into operation flow.Each node can bring into operation the industry of data center
Business system is the business of its distribution.
The method for managing resource of the present embodiment, multiple-node cluster device passes through virtual substrate controller(vBMC)Can be to section
Point is managed collectively, can be according to different resource allocation strategies, flexible configuration resource, reduces the wasting of resources.Additionally, target
The resources such as IO, storage, acceleration, NIC in node can share to all host nodes by operation exchange network.Also, divide
Different types of node(Such as calculate node or memory node), node granularity can be reduced, it is further possible to realize depositing entirely
The density of storage node or full calculate node maximizes configuration.
Embodiment 3
Fig. 3 is the flow chart of the method for managing resource of the embodiment of the present invention three, as shown in figure 3, the area with above-described embodiment
It is not, after the power-up initializing of multiple-node cluster device, if the multiple-node cluster device has the institute of insertion in need
When stating node, the method for managing resource can include:
When step 301, node insertion multiple-node cluster device, hardware signal in place is triggered, SP can be known has node to insert
Enter.
Step 302, for need insertion the node Added Management unit accessory power supply on electricity, wherein, auxiliary tube
The 26S Proteasome Structure and Function for managing unit may refer to the associated description of an embodiment.
Step 303, the service managerZ-HU of the multiple-node cluster device determine the type of the node, if the node is
The host node, then perform step 304, if the node is the destination node, performs step 305.Specifically, service pipe
Reason device can read the electronic labeling information from the memory cell of the Added Management unit of node, according to the electronic tag
Information determines the type of the node.Additionally, service managerZ-HU can also obtain the corresponding groove position information of node, it is follow-up to facilitate
Determine node actually corresponding physical slot position.
Step 304, the service managerZ-HU set up the virtual substrate controller of the node.Examined by virtual substrate controller
Whether normal test node.
Wherein, whether the virtual substrate controller of host node can check the host node normal, can also be according to described
Whether resource allocation strategy inspection has the destination node of membership normal with the host node.
Step 305, the service managerZ-HU complete configuration according to resource allocation strategy.Wherein, service managerZ-HU can match somebody with somebody
The membership and routing iinformation of the host node are put, the routing iinformation is the host node to the industry between other host nodes
Business route.Additionally, service managerZ-HU according to resource allocation strategy, can configure the membership of the destination node.Wherein, such as
Fruit forms operation exchange using the business interface such as insertion multiple-node cluster device such as SAS, SATA interface and PCIe interface backboard
Network.For specific interface type such as PCIe interface, in addition it is also necessary to configure the port type information of each node.
Wherein, the sequential relationship between step 304 and step 305 is not limited, it is also possible to is first carried out step 305 and is performed again
Step 304, or the two execution simultaneously.
After step 304 or step 305, the method for managing resource can also include:
Step 306, service managerZ-HU control electricity on the business main power source of the node.
After step 307, node insertion flow terminate, into operation flow, so that the node can run business.
The method for managing resource of the present embodiment, multiple-node cluster device passes through virtual substrate controller(vBMC)Can be to section
Point is managed collectively, can be according to different resource allocation strategies, flexible configuration resource, reduces the wasting of resources.Additionally, target
The resources such as IO, storage, acceleration, NIC in node can share to all host nodes by operation exchange network.Also, divide
Different types of node(Such as calculate node or memory node), node granularity can be reduced, it is further possible to realize depositing entirely
The density of storage node or full calculate node maximizes configuration.
Embodiment 4
Fig. 4 is the flow chart of the method for managing resource of the embodiment of the present invention four, as shown in figure 4, the area with above-described embodiment
It is not that after the node start to process business, if desired by the knot removal, the method for managing resource can be wrapped
Include:
Step 401, receive knot removal alarm.
Step 402, the operation system of data center discharge the business of the node.
Step 403, service managerZ-HU determine type and groove the position information of the node.If the node is the main section
Point, then perform step 404, if the node is the destination node, performs step 405.
Step 404, service managerZ-HU discharge the virtual substrate controller of the node, and record deletion daily record.
Step 405, will under the node electricity.
Step 406, the node according to the information removing of groove position.
The method for managing resource of the present embodiment, multiple-node cluster device passes through virtual substrate controller(vBMC)Can be to section
Point is managed collectively, can be according to different resource allocation strategies, flexible configuration resource, reduces the wasting of resources.Additionally, target
The resources such as IO, storage, acceleration, NIC in node can share to all host nodes by operation exchange network.Also, divide
Different types of node(Such as calculate node or memory node), node granularity can be reduced, it is further possible to realize depositing entirely
The density of storage node or full calculate node maximizes configuration.
Embodiment 5
Fig. 5 is the structured flowchart of the multiple-node cluster device of the embodiment of the present invention five, as shown in figure 5, the multi-node cluster
Equipment can include:
I/O interface layer 50, for being connected with external equipment;
Operation exchange network 51, is connected respectively with I/O interface layer 50 and mixing resource pool 52;
Mixing resource pool 52 includes multiple nodes, and the node includes host node and destination node, and the host node includes
Processor and internal memory;The destination node includes at least in disk, input-output equipment, network adapter and acceleration component
Plant or various;
Service managerZ-HU 53, distinguishes with I/O interface layer 50, the operation exchange network 51, the resource pool 52 that mixes
Connection, the type for determining the node;If the node is the host node, the virtual substrate control of the node is set up
Device processed, according to resource allocation strategy, configures the membership and routing iinformation of the host node, and the virtual substrate controller is used
There is the destination node of membership in the management host node and with the host node, the routing iinformation is the host node
To the business route between other host nodes;If the node is the destination node, according to resource allocation strategy, institute is configured
State the membership of destination node;Electricity on the business main power source of the node is controlled, so that the node can run business.
Specifically, I/O interface layer 50 can include multiple IO cards, can be connected with the operation system of data center by IO cards
Connect and be adapted to, 10Ge can be supported(10G Ethernets)、IB、GE(Gigabit Ethernet)Accessed Deng legacy network.Operation exchange network
51:Data exchanging function can be provided, various exchange network types are supported, various switching network topologies are supported.Can use many
The form of kind composition operation exchange network, for example:(1)2D/3D Torus, from interconnecting between each node, by straight on backboard
Line road topology composition operation exchange network;(2)Xbar technologies, using interchanger(Switch)Realize the company between each node
Connect, by the composition operation exchange network such as interchanger on backboard;(3)Disk(Disk)It is straight-through, CPU and internal memory is straight with disk
Connect in succession, constitute operation exchange network.The interface type of operation exchange network 51 includes but is not limited to PCIe interface, SAS, SATA
Interface and other self-defined type of network interface.Mixing resource pool 53:Polytype calculating can be provided(CPU+ internal memories), deposit
Storage(Disk), accelerate(FPGA、GPU、DSP), the function such as NIC, IO node, can be using unified system interface and physics chi
Very little specification, supports connection each other.Service managerZ-HU 53:It is responsible for other various pieces in multiple-node cluster device.
The method that each component of multiple-node cluster device carries out resource management, specifically may refer to above-mentioned method for managing resource embodiment
In associated description.
The multiple-node cluster device of the present embodiment, after service managerZ-HU sets up virtual substrate controller for host node, passes through
Virtual substrate controller can be managed collectively to node, can according to different resource allocation strategy, flexible configuration resource,
Reduce the wasting of resources.Additionally, the resource such as IO, storage, acceleration, NIC in destination node can be shared by operation exchange network
To all host nodes.Also, divide different types of node(Such as calculate node or memory node), node granularity can be reduced,
It is further possible to realize that full memory node or the density of full calculate node maximize configuration.
Embodiment 6
Fig. 6 is the structured flowchart of the multiple-node cluster device of the embodiment of the present invention six.Fig. 6 and Fig. 5 identical components have
Identical label, and differring primarily in that for a upper embodiment, as shown in fig. 6, the multiple-node cluster device can also include:
Power supply 61, with I/O interface layer 50, the operation exchange network 51, described mix resource pool 52, the service
Manager 53 is connected respectively, for the multiple-node cluster device initialize during or exist described in insertion in need
It is the Added Management unit of all nodes before the node type that the service managerZ-HU 35 determines the node during node
Upper electricity, the Added Management unit includes accessory power supply, memory cell and interface unit, wherein, preserved in the memory cell
It is host node or the electronic labeling information of destination node, the interface unit and the Service Management to have for identifying the node
Device 53 is connected;
The service managerZ-HU 53 is additionally operable to read the electronic tag from the memory cell of the Added Management unit
Information, the type of the node is determined according to the electronic labeling information.
Wherein, power supply 61 belongs to the basic function component of multiple-node cluster device can provide electricity for multiple-node cluster device
Source, multiple-node cluster device can also be including basic function components such as the fans for being used for radiating.Basic function component can be many
Other various pieces of node cluster device provide the functions such as power supply, radiating.
In a kind of possible implementation method, the service managerZ-HU 53 is additionally operable at the beginning of the multiple-node cluster device
During beginningization, electricity on the business main power source of the destination node is controlled;The electricity on the business main power source of the destination node
Afterwards, electricity on the business main power source of the host node is controlled.In specifically may refer to the method for managing resource of above-described embodiment two
The associated description of initial phase.
In a kind of possible implementation method, the power supply 61 is needed if being additionally operable to the multiple-node cluster device and there are
During the node to be inserted, before the service managerZ-HU 53 determines the node type of the node, to need the institute of insertion
State electricity on the Added Management cell mesh of node.Rank is initialized in the method for managing resource that specifically may refer to above-described embodiment two
The associated description of section.
In a kind of possible implementation method, the host node includes:
Whether the virtual substrate controller is normal for checking the host node;Inspection has with the host node is subordinate to
Whether the destination node of category relation is normal.Rank is initialized in the method for managing resource that specifically may refer to above-described embodiment two, three
Section and node insert the associated description in stage.
In a kind of possible implementation method, the service managerZ-HU 53 is additionally operable in the node start to process business
Afterwards, if desired by the knot removal, notify that the operation system of the data center discharges the business of the node;Determine institute
State type and groove the position information of node;If the node is the host node, the service managerZ-HU 53 discharges the node
Virtual substrate controller, and record deletion daily record will electricity under the node;If the node is the destination node, institute
The record deletion daily record of service managerZ-HU 53 is stated, by electricity under the node;The node according to the information removing of groove position.Specifically may be used
To delete the associated description in stage referring to the method for managing resource interior joint of above-described embodiment four.
In a kind of possible implementation method, multiple-node cluster device also includes backboard(Not shown in figure);
The operation exchange network 51 is connected respectively by the backboard with the I/O interface layer 50, resource pool 52 that mixes
Connect;
The service managerZ-HU 53 is by the backboard and the I/O interface layer 50, the operation exchange network 51, described
Mixing resource pool 52 is connected respectively.
Node in the embodiment of the present invention can have 3.5 inch disk sizes, and the node is connected by unified interface
Device accesses backboard.Referring in Fig. 1 b and its associated description.
The multiple-node cluster device of the present embodiment, after service managerZ-HU sets up virtual substrate controller for host node, passes through
Virtual substrate controller can be managed collectively to node, can according to different resource allocation strategy, flexible configuration resource,
Reduce the wasting of resources.Additionally, the resource such as IO, storage, acceleration, NIC in destination node can be shared by operation exchange network
To all host nodes.Also, divide different types of node(Such as calculate node or memory node), node granularity can be reduced,
It is further possible to realize that full memory node or the density of full calculate node maximize configuration.
The above, specific embodiment only of the invention, but protection scope of the present invention is not limited thereto, and it is any
Those familiar with the art the invention discloses technical scope in, change or replacement can be readily occurred in, should all contain
Cover within protection scope of the present invention.Therefore, protection scope of the present invention described should be defined by scope of the claims.
Claims (11)
1. a kind of method for managing resource, it is characterised in that be applied to the multi-node cluster being connected with the operation system of data center
In equipment, the node of the multiple-node cluster device includes host node Host and destination node Target, and the host node is tool
There is the node of computing capability, the host node includes processor and internal memory;The destination node is the node with storage capacity,
The destination node is including at least one or various in disk, input-output equipment, network adapter and acceleration component;Institute
State the backboard that host node and the destination node insert the multiple-node cluster device by unified interface connector, the main section
Point and the destination node are separate, and methods described includes:
It is the auxiliary of the node during the multiple-node cluster device is initialized or when there is the node of insertion in need
Helping electricity on the accessory power supply of administrative unit, the Added Management unit includes accessory power supply, memory cell and interface unit, its
In, preserved in the memory cell for identifying the electronic labeling information that the node is host node or destination node, it is described
Interface unit is connected with service managerZ-HU;
The service managerZ-HU reads the electronic labeling information from the memory cell of the Added Management unit, according to described
Electronic labeling information determines the type of the node;
If the node is the host node, the service managerZ-HU sets up the virtual substrate controller of the node, according to
Resource allocation strategy, configures the membership and routing iinformation of the host node, and the virtual substrate controller is used to manage institute
State host node and there is the destination node of membership with the host node, the routing iinformation is that the host node is led to other
Business route between node;
If the node is the destination node, the service managerZ-HU configures the target section according to resource allocation strategy
The membership of point;
The service managerZ-HU controls electricity on the business main power source of the node, so that the node can run business.
2. method for managing resource according to claim 1, it is characterised in that in multiple-node cluster device initialization
During, the service managerZ-HU controls electricity on the business main power source of the node, including:
The service managerZ-HU controls electricity on the business main power source of the destination node;
After electricity on the business main power source of the destination node, the service managerZ-HU controls the main electricity of business of the host node
Electricity on source.
3. method for managing resource according to claim 1 and 2, it is characterised in that the service managerZ-HU sets up the section
After the virtual substrate controller of point, including:
Whether the virtual substrate controller checks the host node normal, and according to the resource allocation strategy, inspection and institute
Whether state host node has the destination node of membership normal.
4. method for managing resource according to claim 1 and 2, it is characterised in that the node start to process business it
Afterwards, if desired by the knot removal, methods described also includes:
The operation system of the data center discharges the business of the node;
The service managerZ-HU determines type and groove the position information of the node;
If the node is the host node, the service managerZ-HU discharges the virtual substrate controller of the node, and remembers
Daily record is deleted in record, by electricity under the node;
If the node is the destination node, by electricity under the node;
The node according to the information removing of groove position.
5. method for managing resource according to claim 1 and 2, it is characterised in that
The multiple-node cluster device also includes I/O interface layer and operation exchange network, and the operation exchange network passes through the back of the body
Plate is connected respectively with I/O interface layer, the host node and the destination node;
The service managerZ-HU is by the backboard and I/O interface layer, the operation exchange network, the host node and institute
Destination node is stated to connect respectively.
6. a kind of multiple-node cluster device, it is characterised in that including:
I/O interface layer, for being connected with external equipment;
Operation exchange network, is connected respectively with I/O interface layer and mixing resource pool;
The mixing resource pool includes multiple nodes, and the node includes host node and destination node, the host node be with
The node of computing capability, the host node includes processor and internal memory;The destination node is the node with storage capacity, institute
Destination node is stated including at least one or various in disk, input-output equipment, network adapter and acceleration component;
Backboard, the host node and the destination node insert the back of the body of the multiple-node cluster device by unified interface connector
Plate, the host node and the destination node are separate;
Service managerZ-HU, is connected, for true respectively with I/O interface layer, the operation exchange network, the mixing resource pool
The type of the fixed node;If the node is the host node, the virtual substrate controller of the node is set up, according to money
Source configuration strategy, configures the membership and routing iinformation of the host node, and the virtual substrate controller is used to manage described
Host node and there is the destination node of membership with the host node, the routing iinformation is the host node to other main sections
Business route between point;If the node is the destination node, according to resource allocation strategy, the destination node is configured
Membership;Electricity on the business main power source of the node is controlled, so that the node can run business;
Power supply, connects respectively with I/O interface layer, the operation exchange network, the resource pool, the service managerZ-HU of mixing
Connect, for being initialized in the multiple-node cluster device during or when there is the node of insertion in need, described
It is electricity, the auxiliary tube on the Added Management unit of all nodes before service managerZ-HU determines the node type of the node
Reason unit includes accessory power supply, memory cell and interface unit, wherein, preserved in the memory cell for identifying the section
Point is host node or the electronic labeling information of destination node, and the interface unit is connected with the service managerZ-HU;
The service managerZ-HU is additionally operable to read the electronic labeling information, root from the memory cell of the Added Management unit
The type of the node is determined according to the electronic labeling information.
7. multiple-node cluster device according to claim 6, it is characterised in that the service managerZ-HU, is additionally operable in institute
During stating multiple-node cluster device initialization, electricity on the business main power source of the destination node is controlled;In the target section
On the business main power source of point after electricity, electricity on the business main power source of the host node is controlled.
8. the multiple-node cluster device according to claim 6 or 7, it is characterised in that the host node includes:
Whether the virtual substrate controller is normal for checking the host node;Check to have with the host node and be subordinate to pass
Whether the destination node of system is normal.
9. the multiple-node cluster device according to claim 6 or 7, it is characterised in that the service managerZ-HU, is additionally operable to
After the node start to process business, if desired by the knot removal, notify that the operation system release of data center is described
The business of node;Determine type and groove the position information of the node;If the node is the host node, the Service Management
Device discharges the virtual substrate controller of the node, and record deletion daily record, by electricity under the node;If the node is described
Destination node, then by electricity under the node;The node according to the information removing of groove position.
10. the multiple-node cluster device according to claim 6 or 7, it is characterised in that
The operation exchange network is connected respectively by the backboard with I/O interface layer, the mixing resource pool;
The service managerZ-HU by the backboard and I/O interface layer, the operation exchange network, described mix resource pool
Connect respectively.
11. multiple-node cluster device according to claim 6 or 7, it is characterised in that the node has 3.5 inches of magnetic
Disk size, the node accesses backboard by unified interface connector.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310430685.9A CN103501242B (en) | 2013-09-18 | 2013-09-18 | Method for managing resource and multiple-node cluster device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310430685.9A CN103501242B (en) | 2013-09-18 | 2013-09-18 | Method for managing resource and multiple-node cluster device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103501242A CN103501242A (en) | 2014-01-08 |
CN103501242B true CN103501242B (en) | 2017-06-20 |
Family
ID=49866417
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310430685.9A Active CN103501242B (en) | 2013-09-18 | 2013-09-18 | Method for managing resource and multiple-node cluster device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103501242B (en) |
Families Citing this family (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103955441B (en) * | 2014-04-02 | 2017-04-12 | 华为技术有限公司 | Equipment management system, equipment management method and IO (Input/Output) expansion interface |
CN104123188B (en) * | 2014-06-30 | 2017-10-24 | 华为技术有限公司 | A kind of resource allocation methods and relevant apparatus |
US9934187B2 (en) * | 2014-12-17 | 2018-04-03 | Quanta Computer Inc. | Hot-pluggable computing system |
CN105988908B (en) * | 2015-02-04 | 2018-11-06 | 昆达电脑科技(昆山)有限公司 | The global data processing system of single BMC multiservers |
CN106294062A (en) * | 2016-07-20 | 2017-01-04 | 东南大学 | A kind of method by computing function in server with management function separate design |
CN108228087B (en) * | 2016-12-21 | 2021-08-06 | 伊姆西Ip控股有限责任公司 | Apparatus for hyper-converged infrastructure |
US10571983B2 (en) | 2017-06-28 | 2020-02-25 | International Business Machines Corporation | Continuously available power control system |
US10356008B2 (en) | 2017-06-28 | 2019-07-16 | International Business Machines Corporation | Large scale fabric attached architecture |
US10169048B1 (en) | 2017-06-28 | 2019-01-01 | International Business Machines Corporation | Preparing computer nodes to boot in a multidimensional torus fabric network |
US10088643B1 (en) | 2017-06-28 | 2018-10-02 | International Business Machines Corporation | Multidimensional torus shuffle box |
CN111399973B (en) * | 2019-01-03 | 2023-07-28 | 阿里巴巴集团控股有限公司 | FPGA remote mounting structure, network management and control method and device and electronic equipment |
CN111258763B (en) * | 2020-01-15 | 2024-03-22 | 阿里巴巴集团控股有限公司 | Server system and control method and device thereof |
CN111782232A (en) * | 2020-07-31 | 2020-10-16 | 平安银行股份有限公司 | Cluster deployment method and device, terminal equipment and storage medium |
CN114969149B (en) * | 2022-05-06 | 2024-04-30 | 北京偶数科技有限公司 | Data resource processing method, device and storage medium |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102479099A (en) * | 2010-11-22 | 2012-05-30 | 中兴通讯股份有限公司 | Virtual machine management system and use method thereof |
EP2511822A2 (en) * | 2011-04-15 | 2012-10-17 | Hitachi, Ltd. | Resource management method and computer system |
EP2624138A2 (en) * | 2012-01-31 | 2013-08-07 | VMware, Inc. | Elastic allocation of computing resources to software applications |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
FI20031628A0 (en) * | 2003-11-10 | 2003-11-10 | Nokia Corp | Computer-aided, computer-based unit and method for controlling access to memory between computer-units |
CN102340415B (en) * | 2011-06-23 | 2014-04-16 | 北京新媒传信科技有限公司 | Server cluster system and monitoring method thereof |
-
2013
- 2013-09-18 CN CN201310430685.9A patent/CN103501242B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102479099A (en) * | 2010-11-22 | 2012-05-30 | 中兴通讯股份有限公司 | Virtual machine management system and use method thereof |
EP2511822A2 (en) * | 2011-04-15 | 2012-10-17 | Hitachi, Ltd. | Resource management method and computer system |
EP2624138A2 (en) * | 2012-01-31 | 2013-08-07 | VMware, Inc. | Elastic allocation of computing resources to software applications |
Also Published As
Publication number | Publication date |
---|---|
CN103501242A (en) | 2014-01-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103501242B (en) | Method for managing resource and multiple-node cluster device | |
CN104025063B (en) | Method and apparatus for sharing network interface controller | |
US11615044B2 (en) | Graphics processing unit peer-to-peer arrangements | |
CN110809760B (en) | Resource pool management method and device, resource pool control unit and communication equipment | |
US8145824B1 (en) | Cell-switched inter-partition communications | |
CN108292289B (en) | Method for enabling mini mezzanine Open Computing Project (OCP) plug and play network PHY card | |
CN101983377A (en) | Reserving pci memory space for pci devices | |
US20140188996A1 (en) | Raw fabric interface for server system with virtualized interfaces | |
CN105335168B (en) | Realize system, the method and device of operating system Remote configuration | |
CN108829515A (en) | A kind of cloud platform computing system and its application method | |
CN102473157A (en) | Virtual hot inserting functions in a shared I/O environment | |
CN104881105B (en) | Electronic installation | |
US20200242059A1 (en) | HyperConverged NVMF Storage-NIC Card | |
US20150036681A1 (en) | Pass-through routing at input/output nodes of a cluster server | |
US10158530B2 (en) | Configuration of a cluster server using cellular automata | |
CN104050118B (en) | The devices, systems, and methods of access to functions of the equipments are provided | |
US10657077B2 (en) | HyperConverged NVMF storage-NIC card | |
US11880610B2 (en) | Storage location assignment at a cluster compute server | |
US10088643B1 (en) | Multidimensional torus shuffle box | |
US11055252B1 (en) | Modular hardware acceleration device | |
CN107683593A (en) | Network line card(LC)With host operating system(OS)It is integrated | |
US20210182110A1 (en) | System, board card and electronic device for data accelerated processing | |
CN105005547B (en) | A kind of complete Physical Extents method of multipath server based on NUMA | |
CN103731375B (en) | A kind of FC port virtualization method, device | |
CN109844726A (en) | System and method for distributed AC servo system platform server architecture |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right | ||
TR01 | Transfer of patent right |
Effective date of registration: 20211222 Address after: 450046 Floor 9, building 1, Zhengshang Boya Plaza, Longzihu wisdom Island, Zhengdong New Area, Zhengzhou City, Henan Province Patentee after: Super fusion Digital Technology Co.,Ltd. Address before: 518129 Bantian HUAWEI headquarters office building, Longgang District, Guangdong, Shenzhen Patentee before: HUAWEI TECHNOLOGIES Co.,Ltd. |