WO2022133996A1 - 基于混合云计算平台的存储方法及系统 - Google Patents
基于混合云计算平台的存储方法及系统 Download PDFInfo
- Publication number
- WO2022133996A1 WO2022133996A1 PCT/CN2020/139366 CN2020139366W WO2022133996A1 WO 2022133996 A1 WO2022133996 A1 WO 2022133996A1 CN 2020139366 W CN2020139366 W CN 2020139366W WO 2022133996 A1 WO2022133996 A1 WO 2022133996A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- storage
- data
- computing
- encryption
- scheduling
- Prior art date
Links
- 238000003860 storage Methods 0.000 title claims abstract description 234
- 238000000034 method Methods 0.000 title claims abstract description 25
- 238000012544 monitoring process Methods 0.000 claims abstract description 43
- 230000001133 acceleration Effects 0.000 claims abstract description 39
- 238000012545 processing Methods 0.000 claims abstract description 10
- 230000005540 biological transmission Effects 0.000 claims description 33
- 238000011056 performance test Methods 0.000 claims description 6
- 238000003032 molecular docking Methods 0.000 claims description 4
- 238000012546 transfer Methods 0.000 claims description 4
- 238000005538 encapsulation Methods 0.000 claims description 3
- 238000007726 management method Methods 0.000 description 33
- 238000004590 computer program Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 6
- 238000013500 data storage Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000013523 data management Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000014759 maintenance of location Effects 0.000 description 2
- 239000008186 active pharmaceutical agent Substances 0.000 description 1
- 230000008602 contraction Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000010076 replication Effects 0.000 description 1
- 238000011144 upstream manufacturing Methods 0.000 description 1
Images
Definitions
- the invention relates to the field of computers, in particular to a storage method and system based on a hybrid cloud computing platform.
- the existing cross-cloud storage does not have a unified data management module, and the use of storage objects is strongly bound to the architecture of the corresponding cloud service provider.
- the use of cross-cloud and local storage is inconvenient to use and needs to be specified.
- the existing cross-cloud storage access rate is slow and takes up a lot of computing resources.
- a storage method based on a hybrid cloud computing platform comprising:
- Request judgment Receive user requests and judge whether to store objects or transfer data
- Storage If it is a storage object, create and control the storage object, dynamically monitor and feedback parameters, and create the storage area of the current computing cluster based on the feedback parameters combined with the scheduling strategy;
- Encryption and decryption judgment determine whether to encrypt or decrypt
- Acceleration judgment if it is judged that encryption and decryption is not required, judge whether to accelerate; if it is judged that encryption and decryption is required, after encryption and decryption, judge whether to accelerate;
- Data transmission If no acceleration is required, the data will be transmitted; if the data is uploaded, the data will be acquired and the metadata will be updated to feed back to the user.
- whether there is cross-regional data transmission and dynamic monitoring performance is judged according to the current computing cluster environment, and whether acceleration is required; if it is judged that there is cross-regional data transmission and the dynamic monitoring performance is poor, it is judged that acceleration is required.
- the dynamic performance monitoring includes: monitoring the performance difference of the current computing cluster accessing various public clouds, and returning the obtained performance data and price for recording and statistics.
- the dynamic monitoring includes performing internal regular file upload and download performance tests to determine the performance data of the computing platform accessing each storage moment in each computing cluster to comprehensively analyze the computing platform's scheduling strategy.
- the optimal creation of the storage area of the current computing cluster includes: a scheduling strategy, the scheduling strategy includes: computing cluster scheduling and storage scheduling, and the computing cluster scheduling is based on the computing cluster's bidding model unit price , public cloud vendors bid for the size of the computing resource pool, and whether the computing resource pool satisfies the computation of the computing task; the storage scheduling cooperates with the computing scheduling to allocate storage;
- the creation control process includes: unified storage of resource access rights, storage encryption, and control operations of dynamic monitoring, and the resource information includes regional locations.
- a storage system based on a hybrid cloud computing platform comprising:
- Request judgment unit receive user requests through the access interface module, and judge whether to store objects or transmit data
- Storage unit If it is a storage object, create and control the storage object through the storage management module, and the dynamic monitoring module dynamically monitors and feeds back parameters to the storage management module.
- the most suitable storage area for the current computing cluster If it is a storage object, create and control the storage object through the storage management module, and the dynamic monitoring module dynamically monitors and feeds back parameters to the storage management module. The most suitable storage area for the current computing cluster;
- Obtaining resource information unit in the case of transmission data, obtain resource information such as the regional location of the transmission data through the metadata module;
- Encryption and decryption judgment unit determine whether to encrypt or decrypt
- Acceleration judgment unit if it is judged that encryption and decryption is not required, judge whether to accelerate; if it is judged that encryption and decryption is necessary, perform encryption and decryption processing through the storage encryption module, and after encryption and decryption, judge whether acceleration is necessary; preferably, by combining computing clusters The environment judges whether there is cross-regional data transmission and dynamic monitoring of its poor performance, and judges whether the acceleration module needs to be accelerated;
- Data transmission unit if no acceleration is required, the data is transmitted through the unified storage interface module; if the data is uploaded, the data is acquired and the metadata in the metadata module is updated, and fed back to the user through the access interface module.
- the access interface module is an SDK for encapsulating and docking storage, and is used to transmit user requests and parameters;
- the storage management module a platform storage module based on hybrid cloud bottom storage encapsulation, unified storage control operations, coordinating other modules and strategies of computing clusters in management storage, and performing related scheduling work;
- the unified storage interface module encapsulates all SDKs of public cloud and private cloud, and provides a unified hybrid cloud storage service access interface, so that users can access stored data in the form of file or folder access;
- the metadata management module provides metadata information of the data
- the storage encryption module encrypts and decrypts data
- the dynamic monitoring module used to monitor the performance difference of each public cloud accessed by the current computing cluster, and return the obtained performance data and price to the storage management module for recording and statistics;
- the storage acceleration module accelerates downloading or uploading of data.
- the storage management module performs unified storage control operations including resource access rights, storage encryption, and dynamic monitoring; the resource information includes regional locations.
- the dynamic monitoring module determines the performance data of the computing platform at the moment when each computing cluster accesses each storage by performing a performance test of uploading and downloading files at regular intervals, so that the computing platform can comprehensively analyze the scheduling strategy.
- the scheduling strategy includes: computing cluster scheduling and storage scheduling, the computing cluster scheduling is based on the bidding model unit price of the computing cluster, the size of the bidding computing resource pool of public cloud manufacturers, and whether the computing resource pool satisfies the The operation of the computing task; the scheduling of the storage allocates storage in coordination with the computing scheduling.
- the above-mentioned storage method and system based on the hybrid cloud computing platform selects the storage area of the computing cluster through dynamic monitoring, so as to select the most suitable storage resources in the corresponding scenario, and perform storage based on the hybrid cloud computing platform, enabling cross-cloud storage and local storage.
- the storage is extremely convenient, and the download and upload can be accelerated as needed, so that the cross-cloud storage access rate is fast and saves a lot of time.
- suitable storage objects are automatically allocated according to the storage requirements of the computing platform, and can be quickly searched.
- FIG. 1 is a modular function diagram of a storage system based on a hybrid cloud computing platform according to an embodiment of the present invention
- FIG. 2 is a partial flowchart of a storage method based on a hybrid cloud computing platform according to an embodiment of the present invention.
- the present invention carries out a modular design according to functions, and includes main modules such as a dynamic monitoring module, a storage acceleration module, and a storage management module.
- a storage system based on hybrid cloud storage includes:
- the access interface module is used to transmit user storage requests, including storage creation and permission management;
- the access interface module in this embodiment is an SDK (Software Development Kit) for encapsulating docking storage, and sends the user's request and parameters to the hybrid cloud storage system to complete the user's access operation.
- SDK Software Development Kit
- the storage management module is used to create, manage and allocate storage and other operations. It is equivalent to encapsulating a unified virtual storage based on hybrid storage, and operating based on this unified virtual storage.
- the storage management module in this embodiment is a platform storage module encapsulated based on the underlying storage of the hybrid cloud, which unifies the storage control operations including resource access rights, storage encryption, and dynamic monitoring.
- the storage management module is the central module of the entire hybrid cloud storage. It interacts with other modules through API (Application Programming Interface), and coordinates the strategies of other modules and computing clusters in managing storage to complete related scheduling tasks.
- API Application Programming Interface
- the unified storage interface module is used to provide a unified hybrid cloud storage service access interface. It substantially encapsulates all the SDKs (Software Development Kit) of the public cloud and private cloud, and provides a unified restful api interface through the virtual file system. .
- SDKs Software Development Kit
- the unified storage interface module of this embodiment is used to perform unified interface encapsulation for all public cloud storage interfaces and local slurm cluster storage, so that users can access stored data in the form of file and folder access, and support any two Data replication and synchronization of individual storage objects.
- the metadata management module is used to store metadata information such as data location, upload and update metadata, and download can speed up query efficiency through metadata.
- the metadata management module in this embodiment is used to provide metadata information of data.
- This module is mainly to accelerate the query of data location, size, hash, update time and other information, without the inefficiency caused by public cloud query; and through unified metadata management, users can quickly query the hybrid Which public cloud is the cloud data stored on, and what data is in the corresponding path.
- the storage encryption module is used to encrypt and decrypt data. It mainly supports local storage such as private cloud slurm. It mainly encrypts and decrypts data stored in the local private cloud to ensure data security. This module is optional and can be individually configured for the security level, which will reduce the transmission performance to a certain extent.
- the dynamic monitoring module is used to dynamically monitor the performance difference of the computing cluster accessing each public cloud storage, and return the current rate and price to the storage management module, which is convenient for the storage management module to perform storage scheduling management;
- the dynamic monitoring module in this embodiment is used to monitor the performance difference of the current computing cluster accessing each public cloud, and return the obtained performance data and price to the storage management module for recording and statistics.
- the storage management module for recording and statistics.
- there are performance differences between various public cloud storages, and the network status at different times at home and abroad is also very different.
- the performance data of the computing platform at the moment when each computing cluster accesses each storage is determined, so that the scheduling service of the computing platform can comprehensively analyze the scheduling strategy.
- the storage acceleration module is used to accelerate data download and upload, especially across public cloud storage; for a highly scalable computing cluster, computing nodes need fast elastic expansion and contraction. Costs can also be reduced.
- the storage acceleration module in this embodiment is used to accelerate the downloading and uploading of data from the storage. Since computing tasks are allocated to different hybrid cloud computing clusters according to scheduling policies, and computing clusters may be computing pools distributed around the world, there may be cases where tasks access storage resources across regions. In view of the large difference between domestic and foreign network speeds, a storage acceleration module can be deployed on the proxy node to play a role in CDN acceleration.
- the storage system based on the hybrid cloud storage of the present invention solves the problem of the existing hybrid cloud storage system, especially for the unified scheduling management of storage by the highly scalable computing platform, accelerates the storage upload and download performance, and reduces the overall cost of the hybrid cloud storage.
- An efficient hybrid cloud storage system is provided.
- a storage method based on a hybrid cloud computing platform includes:
- Step S101 request judgment: receive a user request through an access interface module, and judge whether to store an object or transmit data;
- Step S103 storage: if it is a storage object, the storage object is created and controlled by the storage management module, the dynamic monitoring module dynamically monitors and feeds back parameters to the storage management module, and the storage management module combines the scheduling policy with the feedback parameters of the dynamic monitoring module, Create a storage area that is most suitable for the current computing cluster;
- Step S105 obtaining resource information: if it is transmission data, obtain resource information such as the regional location of the transmission data through the metadata module;
- Step S107 encryption and decryption judgment: determine whether to encrypt or decrypt
- Step S109 acceleration judgment: if it is judged that encryption and decryption is not required, then judge whether to accelerate; if it is judged that encryption and decryption is required, perform encryption and decryption processing through the storage encryption module, and after encryption and decryption, judge whether acceleration is required; preferably, by combining The environment of the computing cluster judges whether there is cross-regional data transmission and dynamically monitors its poor performance, and judges whether the acceleration module needs to be accelerated;
- Step S111 data transmission: if no acceleration is required, the data is transmitted through the unified storage interface module; if the data is to be uploaded, the data is acquired and the metadata in the metadata module is updated, and fed back to the user through the access interface module.
- the dynamic monitoring module of this embodiment dynamically monitors the performance difference of the computing cluster accessing various public cloud storages, mainly monitoring the performance of transmission, the size of the transmission file, the time-consuming and rate of uplink time, the time-consuming and rate of downlink time, and the unit price.
- Scheduling strategies include: computing cluster scheduling, storage scheduling.
- the scheduling of computing clusters is mainly based on the bidding model unit price of the computing cluster, the size of the computing resource pool in the bidding of public cloud manufacturers, and whether the computing resource pool satisfies the computing tasks and other strategies.
- Storage scheduling is mainly to allocate appropriate storage in coordination with computing scheduling, and the relative weight is relatively small.
- the storage allocation of computing tasks is based on the area of the public cloud where the computing cluster is located. All areas of alternative storage are determined, and a lower limit of the worst rate is set. Overflow, weight these two parameters with a certain weight, and finally choose the one with the highest weight.
- the dynamic monitoring module needs to deploy sub-processes in each computing cluster. For example, every 10 to 30 minutes, small files of about 10m are regularly executed for uploading and downloading, and the upstream and downstream rates of the nodes at the current time are tested to generate a daily performance table at regular intervals. For example, in the Guangzhou cluster of Cloud A, it took 8s to upload a 10m task to Tencent Cloud Guangzhou cos at 14:00, and the average uplink rate was 1.25m/s, while uploading to the us-east-1 area of AWS took 40s, with an average uplink rate of 40s. Rate 250kb/s. If the lower limit of the worst uplink rate is set to 300kb/s, the storage of aws will pass directly, and the Tencent Cloud Guangzhou cos with better performance will be selected.
- the time rate table can be obtained by calculating the rate of small file transfer. If the rate at most time points is low, such as below the lower threshold of 200kb/s, the performance is judged to be poor.
- the unified storage interface module of this embodiment creates a file object, which includes parameters such as the key, area, storage type, encryption and decryption of each public cloud, etc., so as to be initialized and provided to the user.
- the basic method of adding, deleting, modifying, checking and uploading and downloading is implemented for this file object.
- the bottom layer actually encapsulates the SDKs of major public cloud manufacturers.
- the actual call to this layer of file objects is to call the SDK of the corresponding manufacturer.
- It is a unified file operation layer, which simulates the operation of linux files and encapsulates the SDKs of the major manufacturers at the bottom, and unifies the interface and parameter management.
- the specific process is to create a storage object, that is, a virtual file file system, initialize access rights, encryption and decryption, etc., and then directly use this storage object to upload and download.
- Step S103 in the storage step, if a new storage object is created, the storage object with the highest performance in the current area is allocated to meet the data transmission usage.
- the data is transferred from an existing storage object, and the difference between the real area of the data storage and the current computing cluster area network is poor, you need to transfer the data through the storage acceleration module proxy to meet the extremely high flexibility of computing tasks.
- the acceleration module can be used to speed up the transmission.
- the computing task of the current computing cluster needs to obtain the result data of the previous batch of computing tasks, but it is stored in the us-east-1 object storage of aws, and the current cluster is in the Guangzhou cluster of cloud A, the data transmission rate will be very slow , the time-consuming of data transmission may be greater than the time-consuming of computing tasks, resulting in empty running of computing resources.
- the acceleration module will forward the request to aws s3 through the Singapore proxy deployed in the A cloud public cloud, and the Singapore node connects Guangzhou and North America very quickly.
- Request judgment unit receives user requests through the access interface module, and judges whether to store objects or transmit data
- Storage unit If it is a storage object, create and control the storage object through the storage management module, and the dynamic monitoring module dynamically monitors and feeds back parameters to the storage management module.
- the most suitable storage area for the current computing cluster If it is a storage object, create and control the storage object through the storage management module, and the dynamic monitoring module dynamically monitors and feeds back parameters to the storage management module. The most suitable storage area for the current computing cluster;
- Obtaining resource information unit in the case of transmission data, obtain resource information such as the regional location of the transmission data through the metadata module;
- Encryption and decryption judgment unit determine whether to encrypt or decrypt
- Acceleration judgment unit if it is judged that encryption and decryption is not required, judge whether to accelerate; if it is judged that encryption and decryption is necessary, perform encryption and decryption processing through the storage encryption module, and after encryption and decryption, judge whether acceleration is necessary; preferably, by combining computing clusters The environment judges whether there is cross-regional data transmission and dynamic monitoring of its poor performance, and judges whether the acceleration module needs to be accelerated;
- Data transmission unit if no acceleration is required, the data is transmitted through the unified storage interface module; if the data is uploaded, the data is acquired and the metadata in the metadata module is updated, and fed back to the user through the access interface module.
- the access interface module in this embodiment is an SDK used to encapsulate the docking storage, and is used to transmit user requests and parameters.
- the storage management module of this embodiment based on the platform storage module encapsulated by the underlying storage of the hybrid cloud, unified storage control operations, coordinates other modules and the strategies of the computing cluster in the management storage, and performs related scheduling work;
- the unified storage interface module of this embodiment encapsulates all SDKs of public cloud and private cloud, and provides a unified hybrid cloud storage service access interface, so that users can access stored data in the form of file or folder access;
- the metadata management module of this embodiment provides metadata information of the data
- the storage encryption module of this embodiment encrypts and decrypts data
- the dynamic monitoring module of this embodiment is used to monitor the performance difference of each public cloud accessed by the current computing cluster, and returns the obtained performance data and price to the storage management module for recording and statistics;
- the storage acceleration module of this embodiment accelerates downloading or uploading of data.
- the storage management module unified storage control operations include resource access rights, storage encryption, and dynamic monitoring.
- Resource information includes regional locations.
- the dynamic monitoring module periodically determines the performance data of the computing platform at the moment when each computing cluster accesses each storage by performing a performance test of uploading and downloading files, so that the computing platform can comprehensively analyze the scheduling strategy.
- the scheduling policy includes: computing cluster scheduling and storage scheduling.
- the scheduling of the computing cluster is based on the unit price of the bidding model of the computing cluster, the size of the computing resource pool in the bidding of the public cloud manufacturer, and whether the computing resource pool satisfies the computing tasks.
- Storage scheduling cooperates with computing scheduling to allocate storage.
- the storage method and system based on the hybrid cloud computing platform of the present invention realize unified management and control of storage by integrating the interface APIs of major public cloud manufacturers and common private cloud storage interfaces; according to the storage requirements of the computing platform, appropriate storage is automatically allocated according to the monitoring module object, and can quickly find;
- the acceleration module can quickly complete data processing and save computing costs.
- the embodiments of the present application may be provided as a method, a system, or a computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.
- computer-usable storage media including, but not limited to, disk storage, CD-ROM, optical storage, etc.
- These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory result in an article of manufacture comprising instruction means, the instructions
- the apparatus implements the functions specified in the flow or flow of the flowcharts and/or the block or blocks of the block diagrams.
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
一种基于混合云计算平台的存储方法及系统包括:接收用户请求,判断是存储对象还是传输数据;若为存储对象,则对存储对象进行创建控制处理,动态监控并反馈参数,通过反馈参数择优创建当前计算集群的存储区域;若为传输数据,则获取传输数据的资源信息;判断是否加解密;若判断为不需加解密,则判断是否加速;若判断为需要加解密,进行加解密后,判断是否需要加速;若无需加速则对数据进行传输;若为上传数据则获取数据并更新元数据,反馈给用户;上述方法及系统,通过动态监控择优选择计算集群的存储区域,以在对应的场景下选择最合适的存储资源,根据需要进行加速下载、上传,使跨云存储访问速率快,节约大量时间。
Description
本发明涉及计算机领域,特别涉及一种基于混合云计算平台的存储方法及系统。
随着互联网的发展,数据信息呈现爆炸式增长,特别是随着大数据的来临,给存储系统的进一步发展带来了无限的机遇和挑战。数据存储作为云计算的存储部分,既承担了中间数据的缓存,又是输出有价值数据的重要集中地,对存储系统的管理就非常重要。随着数据量的增大,自建数据中心已不能满足数据存储需求,而使用公有云存储的也越来越多;混合云存储这一使用方式也渐渐被各行各业所使用,但在实际使用中现有混合云存储系统在数据的管理和设计方面存在不足,公有云和私有云存储经常分别独立使用,部署在对应的公有云业务只使用对应公有云的存储,而不能有效的利用跨云存储的资源,无法统一的对所有混合云的资源进行统一管理使用。
而大规模的混合云计算平台面对的实际场景具有计算节点的高伸缩性以及任务调度的分散性,在用户突然提交大批量任务时,计算平台需要在可承受的价格范围内从所有公有云尽可能申请计算资源,按照调度策略将计算任务调度到混合云对应的计算集群里,计算结束后回缩所有的计算节点。这要求在计算节点上对存储传输上花的时间越短越好,以便将计算资源投入到计算中;另一个是调度的随机性,很可能同一批次任务的数据有流水线的依赖关系,但调度的计算云或集群不在一起,需要跨云存储访问数据。
而现有的跨云存储没有统一的数据管理模块,存储对象的使用与对应云服务提供商架构绑定较强,而对跨云以及本地存储的使用存在使用不方便,需要指定的情况。不同的云服务提供商的存储资源的连接速率以及存储成本存在差别,在对应的场景下很难选择最合适的存储资源。现有的跨云存储访问速率较慢,占用了计算资源的大量时间。
发明内容
基于此,有必要提供一种可提高效率的基于混合云计算平台的存储方法。
同时,提供一种可提高效率的基于混合云计算平台的存储系统。
一种基于混合云计算平台的存储方法,包括:
请求判断:接收用户请求,判断是存储对象还是传输数据;
存储:若为存储对象,则对存储对象进行创建控制处理,动态监控并反馈参数,通过反馈参数结合调度策略择优创建当前计算集群的存储区域;
获取资源信息:若为传输数据,则获取传输数据的资源信息;
加解密判断:判断是否加解密;
加速判断:若判断为不需加解密,则判断是否加速;若判断为需要加解密,进行加解密后,判断是否需要加速;
数据传输:若无需加速则对数据进行传输;若为上传数据则获取数据并更新元数据,反馈给用户。
在优选实施例中,根据当前计算集群的环境判断是否有跨区域的数据传输及动态监控性能,判断是否需要加速;若判断有跨区域的数据传输且动态监控性能差,则判断需要加速。
在优选实施例中,所述动态监控性能包括:监控当前计算集群访问各个公有云的性能差异,并将获得的性能数据和价格返回进行记录统计。
在优选实施例中,所述动态监控:通过内部定时对文件进行上传下载的性能测试,确定计算平台在每个计算集群访问每个存储时刻的性能数据,以综合分析计算平台的调度策略。
在优选实施例中,所述优创建当前计算集群的存储区域包括:调度策略,所述调度策略包括:计算集群的调度、存储的调度,所述计算集群的调度根据计算集群的竞价机型单价、公有云厂商竞价计算资源池大小,计算资源池是否满足计算任务的运算;所述存储的调度配合计算调度分配存储;
所述创建控制处理包括:统一存储的资源访问权限、存储加密、动态监控的控制操作,所述资源信息包括区域位置。
一种基于混合云计算平台的存储系统,包括:
请求判断单元:通过访问接口模块接收用户请求,判断是存储对象还是传 输数据;
存储单元:若为存储对象,则通过存储管理模块对存储对象进行创建控制处理,动态监控模块动态监控并反馈参数给存储管理模块,存储管理模块通过动态监控模块的反馈参数结合调度策略,择优创建最适合当前计算集群的存储区域;
获取资源信息单元:若为传输数据,则通过元数据模块获取传输数据的区域位置等资源信息;
加解密判断单元:判断是否加解密;
加速判断单元:若判断为不需加解密,则判断是否加速;若判断为需要加解密,通过存储加密模块进行加解密处理,进行加解密后,判断是否需要加速;优选的,通过结合计算集群的环境判断是否有跨区域的数据传输且动态监控其性能差,判断是否需要加速模块加速;
数据传输单元:若无需加速则通过统一存储接口模块对数据进行传输;若为上传数据则获取数据并更新元数据模块中的元数据,并通过访问接口模块反馈给用户。
在优选实施例中,所述访问接口模块:为用于封装对接存储的SDK,用于传递用户请求和参数;
所述存储管理模块:基于混合云底层存储封装的平台存储模块,统一存储控制操作,在管理存储中协调其他模块以及计算集群的策略,进行相关调度工作;
所述统一存储接口模块:封装公有云和私有云的所有SDK,提供统一的混合云存储服务访问接口,以使用户能以文件或文件夹访问的形式访问存储数据;
所述元数据管理模块:提供数据的元数据信息;
所述存储加密模块:对数据进行加解密;
所述动态监控模块:用于监控当前计算集群访问的各个公有云的性能差异,将得到的性能数据和价格返回给存储管理模块进行记录统计;
所述存储加速模块:对数据进行加速下载或上传。
在优选实施例中,所述存储管理模块统一存储控制操作包括资源访问权限、 存储加密、动态监控;所述资源信息包括区域位置。
在优选实施例中,所述动态监控模块定时通过对文件进行上传下载的性能测试,确定计算平台在每个计算集群访问每个存储的时刻性能数据,以使计算平台的综合分析调度策略。
在优选实施例中,所述调度策略包括:计算集群的调度、存储的调度,所述计算集群的调度根据计算集群的竞价机型单价、公有云厂商竞价计算资源池大小,计算资源池是否满足计算任务的运算;所述存储的调度配合计算调度分配存储。
上述基于混合云计算平台的存储方法及系统,通过动态监控择优选择计算集群的存储区域,以在对应的场景下选择最合适的存储资源,基于混合云计算平台进行存储,使跨云存储及本地存储都极为方便,根据需要进行加速下载、上传,使跨云存储访问速率快,节约大量时间。通过动态监控,针对计算平台的存储需求,自动分配合适的存储对象,并可以快速查找。
图1为本发明一实施例的基于混合云计算平台的存储系统的模块化功能图;
图2为本发明一实施例的基于混合云计算平台的存储方法的部分流程图。
本发明根据功能进行模块化设计,包含如动态监控模块、存储加速模块、以及存储管理模块等主要模块。
如图1所示,本发明一优选实施例的基于混合云存储的存储系统,包括:
访问接口模块,用于传递用户存储请求,包括存储创建和权限管理等;
优选的,本实施例的访问接口模块为一个用于封装对接存储的SDK(Software Development Kit软件开发工具包),将用户的请求和参数发送至混合云存储系统中,完成用户的访问操作。
存储管理模块,用于创建管理分配存储等操作,相当于基于混合存储封装了一个统一的虚拟存储,基于这个统一的虚拟存储进行操作。
优选的,本实施例的存储管理模块为基于混合云底层存储封装的一个平台存储模块,统一了存储的包括资源访问权限、存储加密、动态监控在内的控制 操作。存储管理模块是整个混合云存储的中枢模块,与其他模块都有着API(Application Programming Interface,应用程序接口)交互,并在管理存储中协调其他多个模块以及计算集群的策略,完成相关调度工作。
统一存储接口模块,用于提供统一的混合云存储服务访问接口,实质封装了公有云和私有云的所有SDK(Software Development Kit软件开发工具包),并通过虚拟文件系统对外提供统一的restful api接口。
优选的,本实施例的统一存储接口模块用于针对所有公有云的存储接口以及本地slurm集群存储进行统一的接口封装,让用户能以文件和文件夹访问的形式访问存储数据,并支持任意两个存储对象的数据复制和同步。
元数据管理模块,用于存储数据位置等元数据信息,上传更新元数据,下载可以通过元数据加快查询效率。
优选的,本实施例的元数据管理模块用于提供数据的元数据信息。这个模块主要是为查询数据的位置,大小,哈希,更新时间等信息起到加速作用,而不用再通过公有云查询导致效率低下;并且通过统一的元数据管理,用户可以很快查询到混合云的数据存储在哪个公有云上,对应的路径下有哪些数据等。
存储加密模块,用于为数据进行加解密,主要支持私有云slurm之类的本地存储。主要为本地私有云存储的数据进行加解密,来保障数据安全。该模块为可选项,可针对安全级别做个性化配置,一定程度上会导致传输性能降低。
动态监控模块,用于动态监控计算集群访问各个公有云存储的性能差异,并返回当前速率和价格返回给存储管理模块,方便存储管理模块进行存储调度管理;
优选的,本实施例的动态监控模块用于监控当前计算集群访问各个公有云的性能差异,并将得到的性能数据和价格返回给存储管理模块进行记录统计。目前各个公有云存储之间存在性能差异,而且网络状态在国内外的不同时刻也存在很大差别。通过监控模块内部定时通过小文件进行上传下载的性能测试,确定计算平台在每个计算集群访问每个存储的时刻性能数据,以便计算平台的调度服务综合分析调度策略。
存储加速模块,用于对数据进行加速下载上传,尤其是跨公有云存储;对 于高伸缩计算集群,计算节点需要很快的弹性扩缩容,计算需要的数据下载和上传性能越好,对计算成本也能起到降低的作用。
优选的,本实施例的存储加速模块用于加速数据从存储的下载和上传。由于计算任务会根据调度策略分配在不同的混合云计算集群中,而计算集群可能是分布在全球各地的计算池,其中会存在任务跨区域访问存储资源的情况发生。针对国内外网速差异过大,可在代理节点上部署存储加速模块,起到cdn加速的作用。
本发明的基于混合云存储的存储系统解决了现有混合云存储系统,尤其是针对于高伸缩计算平台对存储的统一调度管理,加速存储上传下载性能,并降低混合云存储的整体成本,实现一套高效的混合云存储系统。
如图2所示,本发明一实施例的基于混合云计算平台的存储方法,包括:
步骤S101,请求判断:通过访问接口模块接收用户请求,判断是存储对象还是传输数据;
步骤S103,存储:若为存储对象,则通过存储管理模块对存储对象进行创建控制处理,动态监控模块动态监控并反馈参数给存储管理模块,存储管理模块通过动态监控模块的反馈参数结合调度策略,择优创建最适合当前计算集群的存储区域;
步骤S105,获取资源信息:若为传输数据,则通过元数据模块获取传输数据的区域位置等资源信息;
步骤S107,加解密判断:判断是否加解密;
步骤S109,加速判断:若判断为不需加解密,则判断是否加速;若判断为需要加解密,通过存储加密模块进行加解密处理,进行加解密后,判断是否需要加速;优选的,通过结合计算集群的环境判断是否有跨区域的数据传输且动态监控其性能差,判断是否需要加速模块加速;
步骤S111,数据传输:若无需加速则通过统一存储接口模块对数据进行传输;若为上传数据则获取数据并更新元数据模块中的元数据,并通过访问接口模块反馈给用户。
进一步,本实施例的动态监控模块动态监控计算集群访问各个公有云存储 的性能差异,主要监控传输的性能,传输文件的大小,上行时间耗时、速率,下行时间耗时、速率,单价。
调度策略包括:计算集群的调度、存储调度。计算集群的调度主要是根据计算集群的竞价机型单价,公有云厂商竞价计算资源池大小,计算资源池是否满足计算任务的运算等策略。存储的调度主要是配合计算调度分配合适的存储,相对起到的权重较小。
择优创建当前计算集群的存储区域,计算任务的存储分配是根据计算集群所在公有云的区域对备选存储所有区域进行判定,设置一个速率最坏下限,性能比较差的直接舍弃,可承受一定单价溢出,对这两个参数有一定的权重加权,最终选择权重最高的。
动态监控模块需要在每个计算集群部署子进程,如每隔10~30min定时执行10m左右的小文件进行上传下载,测试当前时间节点的上行速率和下行速率,生成每日固定间隔时刻性能表。比如A云的广州集群在14:00时测试10m任务上传到腾讯云广州区cos耗时8s,平均上行速率1.25m/s,而上传到aws的us-east-1区耗时40s,平均上行速率250kb/s。如果设置最差上行速率的下限为300kb/s,则aws的存储直接pass,选择性能更好的腾讯云广州区cos。
通过小文件传输计算速率可以得到时间速率表,如果大部分时间点的速率较低,如在下限阀值200kb/s以下,则判断为性能差。
本实施例的统一存储接口模块,创建一个文件对象,包含各个公有云的密钥、区域、存储类型、是否加解密等参数,以便初始化提供给用户。对这个文件对象实现基本的增删改查以及上传下载的方法,底层实际封装各大公有云厂商的sdk,实际调用这一层文件对象,就是调用对应厂商的sdk。即为一个统一的文件操作层,模拟linux文件的操作统一封装底层各大厂商sdk,统一接口和参数管理。具体流程就是创建存储对象,即虚拟文件文件系统,初始化访问权限和加解密等,然后就可以直接用这个存储对象上传下载。
步骤S103,存储步骤中,如果是新建存储对象,则分配当前区域性能最高的存储对象以满足数据的传输使用。但如果是从已存在存储对象中传输数据,且数据存储真实区域与当前计算集群区域网络之间较差,则需要通过存储加速 模块代理传输数据,以满足计算任务极高的弹性。
根据实际使用场景选择是否加速,如果传输速率不能满足快速运算如传输耗时是否超过计算任务耗时,可以通过加速模块加速传输。比如当前计算集群的计算任务需要获取上一批计算任务的结果数据,但是存储在aws的us-east-1对象存储中,而当前集群在A云的广州集群,传输数据的速率就会很慢,可能传输数据的耗时还大于计算任务耗时,导致计算资源空跑。而加速模块就会通过部署在A云公有云的新加坡proxy代理转发请求到aws的s3,而新加坡节点连接广州和北美都很快。
本发明一实施例的基于混合云计算平台的存储系统,还包括:
请求判断单元:通过访问接口模块接收用户请求,判断是存储对象还是传输数据;
存储单元:若为存储对象,则通过存储管理模块对存储对象进行创建控制处理,动态监控模块动态监控并反馈参数给存储管理模块,存储管理模块通过动态监控模块的反馈参数结合调度策略,择优创建最适合当前计算集群的存储区域;
获取资源信息单元:若为传输数据,则通过元数据模块获取传输数据的区域位置等资源信息;
加解密判断单元:判断是否加解密;
加速判断单元:若判断为不需加解密,则判断是否加速;若判断为需要加解密,通过存储加密模块进行加解密处理,进行加解密后,判断是否需要加速;优选的,通过结合计算集群的环境判断是否有跨区域的数据传输且动态监控其性能差,判断是否需要加速模块加速;
数据传输单元:若无需加速则通过统一存储接口模块对数据进行传输;若为上传数据则获取数据并更新元数据模块中的元数据,并通过访问接口模块反馈给用户。
进一步,本实施例的访问接口模块:为用于封装对接存储的SDK,用于传递用户请求和参数。
进一步,本实施例的存储管理模块:基于混合云底层存储封装的平台存储 模块,统一存储控制操作,在管理存储中协调其他模块以及计算集群的策略,进行相关调度工作;
进一步,本实施例的统一存储接口模块:封装公有云和私有云的所有SDK,提供统一的混合云存储服务访问接口,以使用户能以文件或文件夹访问的形式访问存储数据;
进一步,本实施例的元数据管理模块:提供数据的元数据信息;
进一步,本实施例的存储加密模块:对数据进行加解密;
进一步,本实施例的动态监控模块:用于监控当前计算集群访问的各个公有云的性能差异,将得到的性能数据和价格返回给存储管理模块进行记录统计;
进一步,本实施例的存储加速模块:对数据进行加速下载或上传。
在优选实施例中,存储管理模块统一存储控制操作包括资源访问权限、存储加密、动态监控。资源信息包括区域位置。
在优选实施例中,动态监控模块定时通过对文件进行上传下载的性能测试,确定计算平台在每个计算集群访问每个存储的时刻性能数据,以使计算平台的综合分析调度策略。
在优选实施例中,调度策略包括:计算集群的调度、存储的调度。计算集群的调度根据计算集群的竞价机型单价、公有云厂商竞价计算资源池大小,计算资源池是否满足计算任务的运算。存储的调度配合计算调度分配存储。
本发明的基于混合云计算平台的存储方法及系统通过集成各大公有云厂商的接口API以及常用私有云存储接口实现存储的统一管控;针对计算平台的存储需求,按照监控模块自动分配合适的存储对象,并可以快速查找;
针对跨区域的存储访问,通过加速模块可以迅速完成数据处理,节约计算成本。
以上述依据本申请的理想实施例为启示,通过上述的说明内容,相关工作人员完全可以在不偏离本项申请技术思想的范围内,进行多样的变更以及修改。本项申请的技术性范围并不局限于说明书上的内容,必须要根据权利要求范围来确定其技术性范围。
本领域内的技术人员应明白,本申请的实施例可提供为方法、系统、或计算机程序产品。因此,本申请可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且,本申请可采用在一个或多个其中包 含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
本申请是参照根据本申请实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器,使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中,使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品,该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上,使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理,从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。
Claims (10)
- 一种基于混合云计算平台的存储方法,其特征在于,包括:请求判断:接收用户请求,判断是存储对象还是传输数据;存储:若为存储对象,则对存储对象进行创建控制处理,动态监控并反馈参数,通过反馈参数结合调度策略择优创建当前计算集群的存储区域;获取资源信息:若为传输数据,则获取传输数据的资源信息;加解密判断:判断是否加解密;加速判断:若判断为不需加解密,则判断是否加速;若判断为需要加解密,进行加解密后,判断是否需要加速;数据传输:若无需加速则对数据进行传输;若为上传数据则获取数据并更新元数据,反馈给用户。
- 根据权利要求1所述的基于混合云计算平台的存储方法,其特征在于,根据当前计算集群的环境判断是否有跨区域的数据传输及动态监控性能,判断是否需要加速;若判断有跨区域的数据传输且动态监控性能差,则判断需要加速。
- 根据权利要求2所述的基于混合云计算平台的存储方法,其特征在于,所述动态监控性能包括:监控当前计算集群访问各个公有云的性能差异,并将获得的性能数据和价格返回进行记录统计。
- 根据权利要求3所述的基于混合云计算平台的存储方法,其特征在于,所述动态监控:通过内部定时对文件进行上传下载的性能测试,确定计算平台在每个计算集群访问每个存储时刻的性能数据,以综合分析计算平台的调度策略。
- 根据权利要求1至4任意一项所述的基于混合云计算平台的存储方法,其特征在于,所述调度策略包括:计算集群的调度、存储的调度,所述计算集群的调度根据计算集群的竞价机型单价、公有云厂商竞价计算资源池大小,计算资源池是否满足计算任务的运算;所述存储的调度配合计算调度分配存储;所述创建控制处理包括:统一存储的资源访问权限、存储加密、动态监控的控制操作,所述资源信息包括区域位置。
- 一种基于混合云计算平台的存储系统,其特征在于,包括:请求判断单元:通过访问接口模块接收用户请求,判断是存储对象还是传输数据;存储单元:若为存储对象,则通过存储管理模块对存储对象进行创建控制处理,动态监控模块动态监控并反馈参数给存储管理模块,存储管理模块通过动态监控模块的反馈参数结合调度策略,择优创建最适合当前计算集群的存储区域;获取资源信息单元:若为传输数据,则通过元数据模块获取传输数据的区域位置等资源信息;加解密判断单元:判断是否加解密;加速判断单元:若判断为不需加解密,则判断是否加速;若判断为需要加解密,通过存储加密模块进行加解密处理,进行加解密后,判断是否需要加速;数据传输单元:若无需加速则通过统一存储接口模块对数据进行传输;若为上传数据则获取数据并更新元数据模块中的元数据,并通过访问接口模块反馈给用户。
- 根据权利要求6所述的基于混合云计算平台的存储系统,其特征在于,所述访问接口模块:为用于封装对接存储的SDK,用于传递用户请求和参数;所述存储管理模块:基于混合云底层存储封装的平台存储模块,统一存储控制操作,在管理存储中协调其他模块以及计算集群的策略,进行相关调度工作;所述统一存储接口模块:封装公有云和私有云的所有SDK,提供统一的混合云存储服务访问接口,以使用户能以文件或文件夹访问的形式访问存储数据;所述元数据管理模块:提供数据的元数据信息;所述存储加密模块:对数据进行加解密;所述动态监控模块:用于监控当前计算集群访问的各个公有云的性能差异,将得到的性能数据和价格返回给存储管理模块进行记录统计;所述存储加速模块:对数据进行加速下载或上传。
- 根据权利要求7所述的基于混合云计算平台的存储系统,其特征在于,所述存储管理模块统一存储控制操作包括资源访问权限、存储加密、动态监控;所述资源信息包括区域位置;根据当前计算集群的环境判断是否有跨区域的数据传输及动态监控性能,判断是否需要加速;若判断有跨区域的数据传输且动态监控性能差,则判断需要加速。
- 根据权利要求6至8任意一项所述的基于混合云计算平台的存储系统,其特征在于,所述动态监控模块定时通过对文件进行上传下载的性能测试,确定计算平台在每个计算集群访问每个存储的时刻性能数据,以使计算平台的综合分析调度策略。
- 根据权利要求9所述的基于混合云计算平台的存储系统,其特征在于,所述调度策略包括:计算集群的调度、存储的调度,所述计算集群的调度根据计算集群的竞价机型单价、公有云厂商竞价计算资源池大小,计算资源池是否满足计算任务的运算;所述存储的调度配合计算调度分配存储。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/CN2020/139366 WO2022133996A1 (zh) | 2020-12-25 | 2020-12-25 | 基于混合云计算平台的存储方法及系统 |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/CN2020/139366 WO2022133996A1 (zh) | 2020-12-25 | 2020-12-25 | 基于混合云计算平台的存储方法及系统 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022133996A1 true WO2022133996A1 (zh) | 2022-06-30 |
Family
ID=82158671
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2020/139366 WO2022133996A1 (zh) | 2020-12-25 | 2020-12-25 | 基于混合云计算平台的存储方法及系统 |
Country Status (1)
Country | Link |
---|---|
WO (1) | WO2022133996A1 (zh) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115225686A (zh) * | 2022-07-13 | 2022-10-21 | 全通金信控股(广东)有限公司 | 一种基于云存储的代理客户端系统 |
CN115695136A (zh) * | 2022-12-30 | 2023-02-03 | 中国人民解放军军事科学院国防科技创新研究院 | 一种多源数据分布式嵌入式处理装置及其按需配置方法 |
CN116582352A (zh) * | 2023-06-13 | 2023-08-11 | 国网浙江省电力有限公司宁波供电公司 | 一种用于数据交互的智能化数字监管方法及系统 |
CN117033693A (zh) * | 2023-10-08 | 2023-11-10 | 联通沃音乐文化有限公司 | 一种混合模式的云处理的方法及系统 |
CN117828649A (zh) * | 2024-03-06 | 2024-04-05 | 天津帕克耐科技有限公司 | 一种微数据中心系统 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102263825A (zh) * | 2011-08-08 | 2011-11-30 | 浪潮电子信息产业股份有限公司 | 一种基于云位置的混合云存储系统数据传输方法 |
CN103024053A (zh) * | 2012-12-18 | 2013-04-03 | 华为技术有限公司 | 云存储方法、资源调度系统、云存储节点及系统 |
CN108551464A (zh) * | 2018-03-08 | 2018-09-18 | 网宿科技股份有限公司 | 一种混合云的连接建立、数据传输方法、装置和系统 |
US20190155919A1 (en) * | 2017-11-20 | 2019-05-23 | International Business Machines Corporation | Performing deduplication on cloud-resident encrypted data |
-
2020
- 2020-12-25 WO PCT/CN2020/139366 patent/WO2022133996A1/zh active Application Filing
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102263825A (zh) * | 2011-08-08 | 2011-11-30 | 浪潮电子信息产业股份有限公司 | 一种基于云位置的混合云存储系统数据传输方法 |
CN103024053A (zh) * | 2012-12-18 | 2013-04-03 | 华为技术有限公司 | 云存储方法、资源调度系统、云存储节点及系统 |
US20190155919A1 (en) * | 2017-11-20 | 2019-05-23 | International Business Machines Corporation | Performing deduplication on cloud-resident encrypted data |
CN108551464A (zh) * | 2018-03-08 | 2018-09-18 | 网宿科技股份有限公司 | 一种混合云的连接建立、数据传输方法、装置和系统 |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115225686A (zh) * | 2022-07-13 | 2022-10-21 | 全通金信控股(广东)有限公司 | 一种基于云存储的代理客户端系统 |
CN115225686B (zh) * | 2022-07-13 | 2024-08-02 | 全通金信控股(广东)有限公司 | 一种基于云存储的代理客户端系统 |
CN115695136A (zh) * | 2022-12-30 | 2023-02-03 | 中国人民解放军军事科学院国防科技创新研究院 | 一种多源数据分布式嵌入式处理装置及其按需配置方法 |
CN116582352A (zh) * | 2023-06-13 | 2023-08-11 | 国网浙江省电力有限公司宁波供电公司 | 一种用于数据交互的智能化数字监管方法及系统 |
CN116582352B (zh) * | 2023-06-13 | 2023-11-14 | 国网浙江省电力有限公司宁波供电公司 | 一种用于数据交互的智能化数字监管方法及系统 |
CN117033693A (zh) * | 2023-10-08 | 2023-11-10 | 联通沃音乐文化有限公司 | 一种混合模式的云处理的方法及系统 |
CN117033693B (zh) * | 2023-10-08 | 2024-03-08 | 联通沃音乐文化有限公司 | 一种混合模式的云处理的方法及系统 |
CN117828649A (zh) * | 2024-03-06 | 2024-04-05 | 天津帕克耐科技有限公司 | 一种微数据中心系统 |
CN117828649B (zh) * | 2024-03-06 | 2024-05-28 | 天津帕克耐科技有限公司 | 一种微数据中心系统 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2022133996A1 (zh) | 基于混合云计算平台的存储方法及系统 | |
US10048979B2 (en) | Managing virtual machine migration | |
US9703598B2 (en) | Managing virtual machine migration | |
CN116018795A (zh) | 基础设施处理单元 | |
US9720727B1 (en) | Managing virtual machine migration | |
Schmidt et al. | Efficient distribution of virtual machines for cloud computing | |
Moreno-Vozmediano et al. | Iaas cloud architecture: From virtualized datacenters to federated cloud infrastructures | |
CN112579296A (zh) | 基于混合云计算平台的存储方法及系统 | |
CN114666333A (zh) | 一种基于多租户理论面向云计算资源调度问题的控制方法 | |
Convolbo et al. | GEODIS: towards the optimization of data locality-aware job scheduling in geo-distributed data centers | |
CN102981929A (zh) | 磁盘镜像的管理方法和系统 | |
CN103957270A (zh) | 一种云原子单元投递与部署的方法及装置 | |
TW202301118A (zh) | 動態微服務分配機制 | |
Vaquero et al. | Deploying large-scale datasets on-demand in the cloud: treats and tricks on data distribution | |
CN111753326A (zh) | 一种容器云平台云存储资源加密方法、系统及介质 | |
CN110912982A (zh) | 芯片设计云计算系统及方法 | |
George et al. | Hadoop MapReduce for mobile clouds | |
Megino et al. | Using Kubernetes as an ATLAS computing site | |
US10725819B2 (en) | System and method for scheduling and allocating data storage | |
Uta et al. | Scalable in-memory computing | |
Tanimura et al. | Building and evaluation of cloud storage and datasets services on AI and HPC converged infrastructure | |
Peris et al. | Distributed scheduling and data sharing in late-binding overlays | |
Chandra | Effective memory utilization using custom scheduler in kubernetes | |
Viet-DINH | Cloud data management | |
Wilkins-Diehr et al. | Science gateways: Harnessing clouds and software services for science |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 20966561 Country of ref document: EP Kind code of ref document: A1 |