The concept of cloud storage is similar to cloud computing. It refers to a system that gathers various types of storage devices in the network through application software to work together and jointly provide data storage and business access functions through cluster application, grid technology or distributed file system.
In a popular sense, all devices in the cloud storage system are completely transparent to users. Any authorized user can connect with cloud storage through the access network to access cloud storage data. It is a way to consolidate storage applications.
Analysis on the current situation of enterprise storage
Most enterprises strive to meet the growing storage requirements from users and applications by increasing disks, arrays and servers. Over time, the data centers of these enterprises will face the problem of decentralized storage. The data is stored in the disks and systems of the data center and spread all over the enterprise.
The data centralization problem of some systems can be solved through San. However, many sans serve specific application groups. When there are several application groups within the enterprise, there are multiple isolated sans, forming data islands. In this case, many enterprises have introduced the concepts of cloud storage and virtualization into the enterprise’s storage system. How to use the cloud storage mode to integrate existing devices in the existing environment has become the first consideration of many data centers.
Analysis of integration mode
Storage virtualization is a popular way of integration. It integrates multiple target devices or services with other additional functions. Provide comprehensive functional services uniformly. Typical virtualization masks the complexity of the system. Add or integrate new functions, simulate and integrate existing service functions, etc. Virtualization acts on one or more entities, which are used to provide storage resources or services.
Storage virtualization can be realized at three different levels: on the host server based on dedicated volume management software, on the disk array using the firmware of the array controller, or on the storage network using a dedicated virtualization engine.
Host based virtualization
If you only need a single host server (or a single cluster) to access multiple disk arrays, you can use host based storage virtualization technology. The virtualization work is completed on the host server through specific software. The virtualized storage space can span multiple heterogeneous disk arrays.
This kind of virtualization is usually realized by the logical volume management software under the host operating system. Its biggest advantage is the tested stability and openness to heterogeneous storage systems. It exists on the host together with the file system, which is convenient for the close combination of the two to achieve effective storage capacity management. Volumes and file systems can expand or shrink dynamically without downtime.
Storage device based virtualization
When multiple host servers need to access the same disk array, virtualization technology based on array controller can be adopted. At this time, virtualization is completed on the array controller, which divides the storage capacity on an array into multiple storage spaces (LUNs) for access by different host systems.
The intelligent array controller provides data block level consolidation, as well as some additional functions, such as Lun masking, cache, instant snapshot, data replication, etc. With different storage systems, this storage device based virtualization model can optimize performance. This virtualization does not depend on a specific host and can support heterogeneous host systems. However, for each storage subsystem, it is a dedicated and private scheme, which can not cross the restrictions between storage devices, so it can not break the incompatibility between devices.
Storage network based virtualization
The above are all one to many access modes. In the real application environment, many to many access modes are required in many cases, that is, multiple host servers need to access multiple heterogeneous storage devices. The purpose is to optimize resource utilization. One or more users use the same resources, or multiple resources provide services to multiple processes. In this case, the work of storage virtualization needs to be completed on the storage network. This is also a prerequisite for the construction of public storage service facilities. The advantages of the two storage virtualization methods described above can be reflected in storage network virtualization at the same time. It supports data center level storage management and heterogeneous host systems and storage systems.
Many mainstream storage manufacturers use the third method to provide integrated storage virtualization integration solutions for users’ existing storage environment. This method can solve the problems of simplified management interface and data island, but it needs to buy expensive virtual storage devices, which requires a lot of investment in the early stage.
How to make better use of existing equipment and integrate the original environment with less investment is the core issue concerned by many enterprises. Virtualization mainly simplifies management and uses a unified interface to manage all devices, while the interconnection and interworking between underlying devices need to be realized through optical fiber switches. Starting from the network level, multiple isolated fcsans are cascaded to form a large cloud storage environment. Various storage resources exist in the cloud. Users only need a cable to connect to the storage cloud to access the required storage resources. At the same time, the management network of each San is unified in a network domain, and all sans are managed through a single network interface.
Responsible editor: CT