CN106095579B - Container resource allocation methods and device - Google Patents
Container resource allocation methods and device Download PDFInfo
- Publication number
- CN106095579B CN106095579B CN201610416933.8A CN201610416933A CN106095579B CN 106095579 B CN106095579 B CN 106095579B CN 201610416933 A CN201610416933 A CN 201610416933A CN 106095579 B CN106095579 B CN 106095579B
- Authority
- CN
- China
- Prior art keywords
- container
- node
- bandwidth
- resource allocation
- cluster
- 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
- 238000013468 resource allocation Methods 0.000 title claims abstract description 32
- 238000000034 method Methods 0.000 title claims abstract description 24
- 238000004891 communication Methods 0.000 claims abstract description 7
- 238000005516 engineering process Methods 0.000 abstract description 2
- 238000005457 optimization Methods 0.000 description 11
- 238000005192 partition Methods 0.000 description 8
- 230000003068 static effect Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5083—Techniques for rebalancing the load in a distributed system
- G06F9/5088—Techniques for rebalancing the load in a distributed system involving task migration
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The present invention provides a kind of container resource allocation methods and device, belongs to field of communication technology.Container resource allocation methods of the invention, comprising: S1, according to the available bandwidth between the container and container of container N number of in network, using the available bandwidth between two containers as side right value, constructs undirected complete graph of having the right using container as node；S2, according to the complete graph, calculate in each cluster node, the bandwidth between any two node is one the smallest, bandwidth the smallest one between two node any in all clusters is compared, the maximum cluster node of bandwidth is calculated, it wherein, include K node, K≤N in each cluster node；S3, big data is affixed one's name on the cluster node top being calculated, is communicated.
Description
Technical field
The invention belongs to fields of communication technology, and in particular to a kind of container resource allocation methods and device.
Background technique
With universal, the carrying environment that container is replacing virtual machine to run as application of lightweight linux container.Together
When, since the network connection situation of carrying physical host is different, the network connection situation gap between each container is larger, together
When, due to being influenced by the variation of bottom physical machine network condition utilization rate, the network connection situation between each container is also
Dynamic change, but this variation is fluctuated less within the scope of certain time.However, MPI/ is with lightweight linux container
It is universal, container is replacing virtual machine to become the carrying environment of application operation.Simultaneously as the network of carrying physical host connects
Situation difference is connect, the network connection situation gap between each container is larger, simultaneously as by bottom physical machine network condition
The influence of utilization rate variation, the network connection situation between each container is also dynamic change, but this is changed certain
Fluctuation is little in time range.However, the big data application of MPI/Hadoop type, needs to have preferable net between each node
Therefore how network connectivity identifies the preferable one group of container of network connection, becomes the important class for ensureing big data application performance
Topic.
And big data currently is disposed in container in application, relying primarily on the network connectivty choosing between static physical resource
Destination node deployment container is taken, and then in the method for upper vessel portion administration big data system and application, the optimization ginseng that this method is chosen
Number is static, the i.e. bandwidth of physical node.
However, on the one hand, the network occupancy situation of bottom physical resource be it is dynamic, i.e., static 10Mb/s bandwidth is not
The available bandwidth for representing the current physical resource is 10Mb/s, meanwhile, static bandwidth is bigger, available bandwidth also bigger hypothesis from
It is so also invalid.On the other hand, the performance of the physical resource of the bottom of this method concern, and the performance of bottom physical resource is not
It can be simply mapped to the performance of the container resource on upper layer, if bottom physical resource available bandwidth is larger, but due to carrying too
The available bandwidth of more containers, each container may be smaller.
In conclusion being applicable in current method is difficult to select the preferable one group of container of network connection in a period of time,
One group of very poor container of network connectivty is possibly even selected, is answered in the upper vessel portion administration big data that current method selects
With it is low to may cause big data effectiveness, waste of resource, extends the execution time of big data operation.
Summary of the invention
The technical problems to be solved by the invention include asking for above-mentioned existing for existing container resource allocation methods
Topic provides the container resource allocation methods and device of a kind of performance that the big data that optimization is run on container is applied.
Solving technical solution used by present invention problem is a kind of container resource allocation methods, comprising:
S1, according to the available bandwidth between the container and container of container N number of in network, using container as node, with two hold
Available bandwidth between device is side right value, constructs undirected complete graph of having the right；
S2, according to the complete graph, calculate in each cluster node, the bandwidth between any two node is one the smallest,
Bandwidth the smallest one between two node any in all clusters is compared, the maximum cluster node of bandwidth is calculated,
In, it include K node, K≤N in each cluster node；
S3, big data is affixed one's name on the cluster node top being calculated, is communicated.
Preferably, after step s 3 further include:
The available bandwidth between any two container is measured at regular intervals, repeats step S1 and S2.
It may further be preferable that described be at regular intervals measured the available bandwidth between any two container,
It specifically includes:
By IPerf bandwidth measuring unit, the available bandwidth between any two container is surveyed at regular intervals
It is fixed.
Preferably, band is calculated in described be compared bandwidth the smallest one between two node any in all clusters
A wide maximum cluster node, specifically includes:
Using heuristic greedy principle, the smallest middle bandwidth of bandwidth between any two node is calculated in all clusters
A maximum cluster node.
It may further be preferable that the inspiration condition of the heuristic greedy principle is the bottleneck figure based on the complete graph
Data structure.
Solving technical solution used by present invention problem is a kind of container resource allocation device, comprising:
Complete graph constructing module, for the available bandwidth between the container and container according to container N number of in network, with container
Undirected complete graph of having the right is constructed using the available bandwidth between two containers as side right value for node；
Optimal cluster node obtains module, for calculating in each cluster node according to the complete graph, any two node it
Between bandwidth it is one the smallest, bandwidth the smallest one between two node any in all clusters is compared, bandwidth is calculated
A maximum cluster node, wherein include K node, K≤N in each cluster node；
Communication module is communicated for affixing one's name to big data on the cluster node top being calculated.
Preferably, the container resource allocation device further include: bandwidth measuring unit, it is right at regular intervals to be used for
Available bandwidth between any two container is measured.
It may further be preferable that the bandwidth measuring unit includes IPerf bandwidth measuring unit.
Preferably, the optimal cluster node obtains module and is specifically used for, and using heuristic greedy principle, institute is calculated
There is in cluster the bandwidth the smallest one middle maximum cluster node of bandwidth between any two node.
It may further be preferable that the inspiration condition of the heuristic greedy principle is the bottleneck figure based on the complete graph
Data structure.
The invention has the following beneficial effects:
Container resource allocation device provided by the present invention can identify the preferable cluster of connectivity in all containers
Container node, with heuristic feasible operator principle ensure the aggregate bandwidth between the cluster inner pressurd vessel maximum, and the container across cluster it
Between bandwidth it is minimum, so that the container preferentially chosen with group disposes big data application, so optimize run on it is big on container
The performance of data application.
Detailed description of the invention
Fig. 1 is the flow chart of 1 container resource allocation methods of the embodiment of the present invention；
Fig. 2 is by the step S1 of 1 container resource allocation methods of the embodiment of the present invention undirected complete graph of having the right constructed
Schematic diagram；
Fig. 3 is the schematic diagram of 1 container resource allocation device of the embodiment of the present invention.
Specific embodiment
Technical solution in order to enable those skilled in the art to better understand the present invention, with reference to the accompanying drawing and specific embodiment party
Present invention is further described in detail for formula.
Embodiment 1:
As shown in Figure 1, including the following steps: the present embodiment provides a kind of container resource allocation methods
S1, according to the available bandwidth between the container and container of container N number of in network, using container as node, with two hold
Available bandwidth between device is side right value, constructs undirected complete graph of having the right.
As shown in Fig. 2, each container can regard a node in figure as.Due to container connection between any two, and two
Communication between container is not necessarily to forward via other containers, then the network topology of whole containers can regard as one it is undirected completely
Figure, meanwhile, the bandwidth between any two container node can be by IPerf equiband measuring unit results of regular determination, and then gives
Statistics available bandwidth in certain time interval out.The weight setting on side in Fig. 1 can be counted available bandwidth thus, such as connection section
The weight on the side (u, v) of point u and node v is the statistics available bandwidth between container node u and v, is indicated with b ω (u, v), then
The problem of finding connectivity preferable one group of container can be converted into the Graph partition problem of complete graph.
S2, according to the complete graph, calculate in each cluster node, the bandwidth between any two node is one the smallest,
Bandwidth the smallest one between two node any in all clusters is compared, the maximum cluster node of bandwidth is calculated,
In, it include K node, K≤N in each cluster node.
The step, cluster container node refer to the combination of multiple containers node；By taking the container node of N=10 i.e. 10 as an example,
If necessary to the node cluster interior joint number K=3 of selection, then the node cluster to be chosen of the application is that 3 are chosen from 10 nodes
One of all possibilities of node, it is known that it may select to be 10x9x8=720 kind, that is, what is chosen is possible in 720
1 kind in scheme.Specifically, in step sl according to the constructing network topology of container one undirected complete graph of having the right, therefore,
Container node optimization sub-clustering problem can be converted to the weight size that one connects side according to node and carry out the excellent of complete graph division
Change problem, optimization aim is that the bandwidth between the container node ensured in same node cluster is as big as possible, if being at most divided into k
A network cluster, node cluster can use Ci, i=1...k expression, then the target equation of optimization problem may be defined as formula (1).
max(bω(u,v))u,v∈Ci, i=1...k (1)
Simultaneously it is desirable that the bandwidth resources between each container node cluster are evenly distributed, i.e., to all container node clusters and
Speech maximizes the minimum value of bandwidth between any two container node in cluster, then target equation is shown in formula (2).
Wherein, container node optimization sub-clustering problem (Optimized Virt.Nodes k-Clustering problem)
Optimize sub-clustering problem as container node using the container node sub-clustering problem that formula (2) is optimization aim equation.
For convenience of explanation, the present invention introduces a classical problem in complete graph Research on partition field in graph theory, and minimum is most
The definition of big k cluster partition problem.
Minimax k cluster partition problem (min-max k-Clustering problem) given one have the right it is undirected completely
Figure, is shown in formula (3),
Minimax k cluster partition problem, which is defined as dividing G=(V, G), to be at most k cluster and makes appointing in any cluster
The maximum value for the weight on side between two nodes of anticipating is minimum, i.e. hypothesis C1....CkIt is ready-portioned cluster, then optimization aim equation is shown in formula
(4)。
Wherein, container node optimization sub-clustering problem and minimax k cluster partition problem are equivalence problems.
It proves: enabling
Then formula (4) is converted into formula (6):
It enables
F (b ω (u, v))=b ω (u, v), b ω (u, v) > 0, u, v ∈ Ci, i=1...k (7)
According to the mathematical optimization problem definition of standard, formula (2) can be write
According to the mathematical optimization problem definition of standard, formula be can be written as
Formula (11) can be obtained by formula (7) and formula (8) and know that f (b ω (u, v)) and g (b ω (u, v)) change with b ω (u, v)
Trend is on the contrary, so formula
Max (f (b ω (u, v)))=min (g (b ω (u, v))), b ω (u, v) > 0 (11)
So formula (9) and (10) are of equal value, i.e., formula (2) and formula (6) are of equal value.
However, minimax k cluster partition problem is NP-Hard problem, only when the weight on side between node meets triangle
When inequality, i.e., formula (12) meets
Just there is the approximate algorithm for determining the approximation coefficient upper bound in minimax k cluster partition problem, i.e. the problem is that APX is asked
Topic.
Bandwidth the smallest one between two node any in all clusters is compared and bandwidth is calculated most by the present embodiment
A big cluster node is calculated in all clusters that bandwidth is the smallest between any two node specifically using heuristic greedy principle
One middle maximum cluster node of bandwidth.Namely a kind of heuritic approach solving virtual resource node is devised in the present embodiment
Optimize sub-clustering problem, inspiration condition is that the bottleneck figure based on complete graph finds maximum independent set, is then with these independent sets
Central configuration node cluster.
Wherein, bottleneck figure (bottleneck graph) gives specific weight values ω, the bottleneck of an authorized graph G=(V, G)
Figure is defined as formula (13).
WhereE (ω)={ e ∈ E: ω (e)≤ω } (13) Gb (ω)=G (V, (ω))
The bottleneck figure Gb (ω) that side of all weights greater than ω is formed original image G is deleted from original image, for side
In the case that weight is positive value, Gb (0) is exactly the vertex set of original image G, and side collection is sky；Gb(ωmax) it is exactly original image G itself.
S3, big data is affixed one's name on the cluster node top being calculated, is communicated.
It after step s 3, can also include being surveyed at regular intervals to the available bandwidth between any two container
Fixed step continues to repeat step S1 and S2 later；It wherein, specifically can be by IPerf bandwidth measuring unit, every certain
Time is measured the available bandwidth between any two container.Wherein, time interval please can specifically be limited according to specifically,
And why to be measured at regular intervals to the available bandwidth between any two container is because the occupancy of bandwidth is more
It is few different time according to user number and it is different, therefore this kind of mode may insure that container resource allocation is optimal.
In conclusion container resource allocation methods provided by the present embodiment can identify connectivity in all containers
Preferable cluster container node ensures that the aggregate bandwidth between the cluster inner pressurd vessel is maximum with heuristic feasible operator principle, and across
Bandwidth between the container of cluster is minimum, to preferentially choose the container deployment big data application with group, and then optimizes and runs on appearance
The performance of big data application on device.
Embodiment 2:
As shown in figure 3, the container money in embodiment 1 can be used the present embodiment provides a kind of container resource allocation device
Source distribution method.Container resource allocation device in the present embodiment includes: complete graph constructing module, optimal cluster node acquisition mould
Block, communication module；Wherein, complete graph constructing module, for the available band between the container and container according to container N number of in network
Width, using the available bandwidth between two containers as side right value, constructs undirected complete graph of having the right using container as node；Optimal cluster knot
Point obtains module, for calculating in each cluster node, the bandwidth between any two node is the smallest by one according to the complete graph
Bandwidth the smallest one between two node any in all clusters is compared and the maximum cluster node of bandwidth is calculated by item,
It wherein, include K node, K≤N in each cluster node；Communication module, in the big number of cluster node top administration being calculated
According to being communicated.
Wherein, the optimal cluster node obtains module and is specifically used for, and using heuristic greedy principle, all clusters are calculated
In the bandwidth the smallest one middle maximum cluster node of bandwidth between any two node.The inspiration item of the heuristic greedy principle
Part is the data structure of the bottleneck figure based on the complete graph.
Certainly, the device resource allocation device in the present embodiment can also include: bandwidth measuring unit, be used for every certain
Time is measured the available bandwidth between any two container.Wherein, bandwidth measuring unit includes IPerf bandwidth measurement list
Member.
In conclusion container resource allocation device provided by the present embodiment can identify connectivity in all containers
Preferable cluster container node ensures that the aggregate bandwidth between the cluster inner pressurd vessel is maximum with heuristic feasible operator principle, and across
Bandwidth between the container of cluster is minimum, to preferentially choose the container deployment big data application with group, and then optimizes and runs on appearance
The performance of big data application on device.
It is understood that the principle that embodiment of above is intended to be merely illustrative of the present and the exemplary implementation that uses
Mode, however the present invention is not limited thereto.For those skilled in the art, essence of the invention is not being departed from
In the case where mind and essence, various changes and modifications can be made therein, these variations and modifications are also considered as protection scope of the present invention.
Claims (10)
1. a kind of container resource allocation methods characterized by comprising
S1, according to the available bandwidth between the container and container of container N number of in network, using container as node, with two containers it
Between available bandwidth be side right value, construct undirected complete graph of having the right；
S2, according to the complete graph, calculate in each cluster node, the bandwidth between any two node is one the smallest, by institute
There is in cluster between any two node bandwidth the smallest one to be compared calculating, obtain cluster where maximum one, wherein
It include K node, K≤N in each cluster node；
S3, big data is affixed one's name on the cluster node top being calculated, is communicated.
2. container resource allocation methods according to claim 1, which is characterized in that after step s 3 further include:
The available bandwidth between any two container is measured at regular intervals, repeats step S1 and S2.
3. container resource allocation methods according to claim 2, which is characterized in that described at regular intervals to any two
Available bandwidth between a container is measured, and is specifically included:
By IPerf bandwidth measuring unit, the available bandwidth between any two container is measured at regular intervals.
4. container resource allocation methods according to claim 1, which is characterized in that described by two node any in all clusters
Between bandwidth the smallest one be compared the maximum cluster node of bandwidth be calculated, specifically include:
Using heuristic greedy principle, it is maximum that bandwidth the smallest one middle bandwidth between any two node is calculated in all clusters
A cluster node.
5. container resource allocation methods according to claim 4, which is characterized in that the inspiration of the heuristic greedy principle
Condition is the data structure of the bottleneck figure based on the complete graph.
6. a kind of container resource allocation device characterized by comprising
Complete graph constructing module is section with container for the available bandwidth between the container and container according to container N number of in network
Point constructs undirected complete graph of having the right using the available bandwidth between two containers as side right value；
Optimal cluster node obtains module, for calculating in each cluster node, between any two node according to the complete graph
Bandwidth is one the smallest, is compared calculating for bandwidth the smallest one between two node any in all clusters, obtains wherein most
Cluster where one of big, wherein include K node, K≤N in each cluster node；
Communication module is communicated for affixing one's name to big data on the cluster node top being calculated.
7. container resource allocation device according to claim 6, which is characterized in that the container resource allocation device also wraps
Include: bandwidth measuring unit is used at regular intervals be measured the available bandwidth between any two container.
8. container resource allocation device according to claim 7, which is characterized in that the bandwidth measuring unit includes
IPerf bandwidth measuring unit.
9. container resource allocation device according to claim 6, which is characterized in that the optimal cluster node obtains module tool
Body is used for, and using heuristic greedy principle, the smallest middle bandwidth of bandwidth between any two node is calculated in all clusters
A maximum cluster node.
10. container resource allocation device according to claim 9, which is characterized in that the heuristic greedy principle opens
Clockwork spring part is the data structure of the bottleneck figure based on the complete graph.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610416933.8A CN106095579B (en) | 2016-06-14 | 2016-06-14 | Container resource allocation methods and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610416933.8A CN106095579B (en) | 2016-06-14 | 2016-06-14 | Container resource allocation methods and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106095579A CN106095579A (en) | 2016-11-09 |
| CN106095579B true CN106095579B (en) | 2019-06-07 |
Family
ID=57846759
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610416933.8A Active CN106095579B (en) | 2016-06-14 | 2016-06-14 | Container resource allocation methods and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106095579B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106911777A (en) * | 2017-02-24 | 2017-06-30 | 郑州云海信息技术有限公司 | A kind of data processing method and server |
| CN109525407B (en) * | 2017-09-18 | 2020-05-26 | 中国科学院声学研究所 | Method for generating same-layer intersection-free full-coverage nested container and readable storage medium |
| CN109491790B (en) * | 2018-11-02 | 2021-08-27 | 中山大学 | Container-based industrial Internet of things edge computing resource allocation method and system |
| CN111273871B (en) * | 2020-01-19 | 2021-05-04 | 星辰天合（北京）数据科技有限公司 | Method and device for dynamically allocating storage resources on container platform |
| CN112104725B (en) * | 2020-09-09 | 2022-05-27 | 中国联合网络通信集团有限公司 | Container mirror image duplicate removal method, system, computer equipment and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104507111A (en) * | 2014-12-26 | 2015-04-08 | 清华大学 | Cluster-based cooperative communication method and device in cellular network |
| CN104636204A (en) * | 2014-12-04 | 2015-05-20 | 中国联合网络通信集团有限公司 | Task scheduling method and device |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8909602B2 (en) * | 2011-02-03 | 2014-12-09 | Vmware, Inc. | Programmatic snapshot and revert of virtualized data center inventory |
| US8813088B2 (en) * | 2011-05-26 | 2014-08-19 | International Business Machines Corporation | Scheduling flows in a multi-platform cluster environment |
-
2016
- 2016-06-14 CN CN201610416933.8A patent/CN106095579B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104636204A (en) * | 2014-12-04 | 2015-05-20 | 中国联合网络通信集团有限公司 | Task scheduling method and device |
| CN104507111A (en) * | 2014-12-26 | 2015-04-08 | 清华大学 | Cluster-based cooperative communication method and device in cellular network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106095579A (en) | 2016-11-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106095579B (en) | Container resource allocation methods and device | |
| US20220209971A1 (en) | Methods and apparatus to aggregate telemetry data in an edge environment | |
| US11704144B2 (en) | Creating virtual machine groups based on request | |
| US11018979B2 (en) | System and method for network slicing for service-oriented networks | |
| US10200261B2 (en) | Multiple-computing-node system job node selection | |
| CN107196869B (en) | The adaptive load balancing method, apparatus and system of Intrusion Detection based on host actual loading | |
| CN109032801B (en) | Request scheduling method, system, electronic equipment and storage medium | |
| Chen et al. | Energy-efficient fault-tolerant data storage and processing in mobile cloud | |
| Mechalikh et al. | PureEdgeSim: A simulation framework for performance evaluation of cloud, edge and mist computing environments | |
| Zhu et al. | BLOT: Bandit learning-based offloading of tasks in fog-enabled networks | |
| US20130263151A1 (en) | Consistent Hashing Table for Workload Distribution | |
| US10621002B2 (en) | Iterative task centric resource scheduling for a user program between different computing frameworks | |
| US10027596B1 (en) | Hierarchical mapping of applications, services and resources for enhanced orchestration in converged infrastructure | |
| CN108475207A (en) | The joint auto zoom of cloud application | |
| CN109947551A (en) | A kind of more round method for allocating tasks, edge calculations system and its storage medium | |
| CN106790332B (en) | Resource scheduling method, system and main node | |
| US20230119552A1 (en) | Resource management mechanisms for stateful serverless clusters in edge computing | |
| EP4300310A1 (en) | Opportunistic placement of compute in an edge network | |
| Li et al. | Resource scheduling based on improved spectral clustering algorithm in edge computing | |
| Ferrer et al. | Ad-hoc edge cloud: A framework for dynamic creation of edge computing infrastructures | |
| Dinh et al. | An energy efficient integration model for sensor cloud systems | |
| Yin et al. | Joint scheduling of data and computation in geo-distributed cloud systems | |
| CN110532060A (en) | A kind of hybrid network environmental data collecting method and system | |
| US20210319281A1 (en) | Subtask Assignment for an Artificial Intelligence Task | |
| Fukunaga et al. | Virtual machine placement for minimizing connection cost in data center networks |
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 |