什么是多云存储

The cloud and cloud storage are essentially synonymous. So, while on-premise storage means data storage housed and managed at a company’s physical location, storing data “in the cloud” fundamentally is the same as storing it at an off-site data center accessible by internet. The off-site cloud storage is most typically a third-party cloud service provider that could offer both managed and self-service features.

Managed features allows companies to out-source their IT needs to cloud providers, effectively reducing that departmental functionality to a budget line item. Highly visible examples of these services are SaaS (Software-as-a-Service) vendors, who manage the entire back-end and user interface for clients that simply pay for the service and may never even think twice about security, or data management. Likewise, IaaS (Infrastructure-as-a-Service) provides infrastructure, like servers, CPU, GPU, RAM, storage, network, under managed and self-service models.

Self-service models, or unmanaged models, are bare services that provide intrepid developers the basics for installing and self-managing their own environments. Self-management infrastructure and services are good for consumers who have the expertise, or require custom configurations. This approach will undeniably reduce CSP costs, but is offset with zero management or software support.

Cloud storage has many properties that make it a flexible solution that has become a source of innovation and advantage for many organizations. Utilizing cloud storage has allowed companies to easily comply with regulations and secure their data assets, as well as ensure they are adequately protected in the case of data disasters and loss. Developers can easily move in and out of environments as they experiment with workloads, as readily as enterprises can scale their data stores to meet the demand of their users.

Cloud management platforms, or cloud data controllers, provide a global view of cloud resources, with controls to manage data between clouds. Generally, these applications are standalone solutions that create a namespace that spans both on-premise, and cloud resources, with features for automation and orchestration of data tasks, that help to maintain data integrity, protection, and security, for each data storage environment.

• Single Namespace — Cloud platforms create global file systems that span multiple cloud spaces. These differ from cloud migration tools that simply migrate data between environments, but leave the separate management of each cloud storage pool to IT. A single namespace allows the incorporation of on-premise stores with multiple clouds, searchable through metadata searches across public cloud and private clouds.
• Cloud Native Storage — The cloud data controller should accommodate storing data in formats native to the cloud storage provider, that is, data stored in Microsoft’s Azure Blob storage native format should be stored in native Azure Blob format, and if later copied to Amazon S3, it should be converted into native S3 format. Native formats help avoid requiring additional drivers or software, and leverage the compute resources native to those clouds.
• Standalone Solution — Technically, cloud platforms are standalone solutions, but they may be sold as an add-on solution to a vendor’s core storage offering intended as the primary purchase. While bundles can be good, ensure that the offer matches business goal needs. These multicloud add-ons may fall short in features compared to other standalone solutions, moreover, they may precipitate vendor lock-in.
• Automation and Orchestration — Universality is a key feature of best-in-class multicloud solutions, which is achieved through provision of an API set with a workflow engine that can be scripted to perform automations, and more complex orchestrations between cloud resources. Automations and orchestrations enable many of the benefits of implementing multicloud strategies.

Multicloud platforms are designed to remove many of the obstacles to configuring, monitoring, and managing multiple cloud storage services together. Common challenges include:

• Cloud Integrations and Multiple APIs — Cloud integrations require the use of APIs to communicate seamlessly. While this is possible, cloud service providers are far from standardized, potentially creating misalignments between the API methodologies of disparate vendors. Carefully evaluate the features of cloud management platforms, which can offer solutions for API management, to ensure they are standards-based, and support necessary cloud service providers.
• Privacy and Security — Any multicloud storage strategy must consider data privacy and security. While cloud storage does provide a greater threat surface over traditional on-premise network security, cloud security can be equally if not more secure, but must be managed. All resources within the cloud storage fabric must be aligned with overarching security policies, data must be protected through encryption at rest and in transit, and user access must be authenticated and authorized.
• Inherent Data Network Latency — Data latency is a physical limitation of networks, which multicloud, through strategic positioning of cloud resources, attempts to overcome. Access to data anywhere, anytime, easily and quickly, becomes a greater challenge as more services are added to a company’s cloud storage fabric, and more data is communicated between services. While this is a challenge that can be overcome with forethought, it must be understood and planned for.

Multicloud storage provides strategic business benefits in several use cases that aim to enhance and bolster data protection, security, flexibility, performance, accessibility, and resilience. In short, cases using multicloud aim to ensure users have access to their data, anywhere, anytime, with minimal lag, delivered to them securely.

• Clustering Mission-Critical Databases — This strategy exploits the advantage of using multicloud storage to isolate specific functional databases clusters in the cloud. This achieves two things, frees the functional database from block storage typical in on-premise data storage, and allows the database to be duplicated across multiple clouds. This allows data to migrate around to locations as needed, to be replicated and serve peak load times, and provide a shielding of mission critical data against downtimes.
• Easing Disaster Recovery — Applications are also data. By distributing both data and apps that support business operations to multiple clouds, the same diversification mechanism can be used for disaster recovery plans. If systems are compromised, and the network goes offline, redundant recovery stores can help organizations rapidly bring services back up.
• Backup and Archiving — A fundamental use of multicloud storage is to provide a reliable off-site repository for data backups and archives. Off-site replication can be less expensive, easier, and more reliable than on-premise archiving.
• Resilience and Performance Strategies — Multiple data stores in different geographic locations is a key risk mitigation strategy. If one location experiences downtime, other locations can take up the workload until the system returns to normal. This approach can be pushed even further by diversifying vendors, in addition to diversifying the geographic location of data centers. Multiple vendors can balance workloads, in the case that one vendor services begins to underperform, others can take up workloads in the local area.
• Compliance Strategies — Data carries with it regulatory requirements, and organizations dealing in data must guarantee and show their regulatory compliance. The general use case is to store regulated data, like personally identifiable information, or health information, on data storage that is physically located within the sovereign boundary of the country of regulation.