SNIA (Storage Network Industry Association), an international authoritative organization in the storage industry, gave the definition of Storage Virtualization: "by abstracting, hiding, or hiding internal functions of storage systems/subsystems from applications, computing servers, and network resources. Isolation, independent of application, network storage and data management." The storage virtualization technology abstracts the underlying storage devices in an abstract manner and shields the server layer from the uniqueness of the storage device hardware, while retaining only its unified logical features, thereby achieving centralized, unified, and convenient management of the storage system.
Compared with traditional storage, the advantages of virtualized storage are mainly reflected in: high disk utilization, disk utilization of traditional storage technologies is generally only 30-70%, and disk utilization after virtualization technology is as high as 70-90%; Flexible storage, which can adapt to different vendors and different types of heterogeneous storage platforms, providing better flexibility for storage resource management; convenient management, providing a centralized management method for large-capacity storage systems, and avoiding the expansion of storage devices. The management troubles are caused. With better performance, the virtualized storage system can perform load balancing. The bandwidth required for each data access is allocated to each storage module and the overall access bandwidth of the system is improved.
There are multiple classification methods for virtual storage, which can be divided into two major categories: classification based on the location of virtualization in the I/O path; classification based on different control paths and data paths. According to different locations where virtualization is implemented in the I/O path, virtualized storage can be divided into host virtual storage, network virtual storage, and storage device virtual storage. According to different control paths and data paths, virtualized storage is divided into symmetrical virtualization and asymmetric virtualization.
The system of virtualized storage constitutes a typical storage method in cloud storage as distributed storage. In this way, the storage device is generally managed in an out-of-band virtualization manner. Metadata management and data transmission are accomplished through an IP network. There are four main types of storage devices in this type of virtualized storage system.
Client:
Clients provide various application services to customers, such as web services, databases, file services, scientific computing, etc. The storage agent software runs on the client to provide read and write access to applications for network virtual appliance provisioning.
Configure the management server:
The configuration management server is used to configure and manage the system. Log in to the cloud storage platform through the Internet, Telnet, or other interface to remotely configure and manage the entire storage system.
Metadata server:
The cloud storage system's metadata server (MDS) manages the layout information of the entire system's metadata and object data, and is responsible for system resource allocation and network virtual disk address mapping. Globally virtualized storage management software and cluster management software deployed on the MDS can manage the configuration and operation of the entire storage system. In addition, the MDS implements redundant data relationships among common storage nodes through redundant management software.
Object storage node:
Each storage node is an independent storage device that is responsible for the storage, backup, migration, and recovery of object data, and is responsible for monitoring the operating status and resource status of other storage devices. At the same time, the storage node runs virtualized storage management software and stores the data needed by the application.
The three-tier model of virtualized storage According to the composition and characteristics of cloud storage systems, virtualized storage models can be divided into three layers: physical device virtualization layer, storage node virtualization layer, storage area network virtualization layer.
The three-tier virtualization storage model greatly reduces the complexity of storage management, effectively encapsulates the complexity and diversity of the underlying storage devices, and enables the system to have better scalability and flexibility. Users can create virtual volumes simply by adding the storage devices to the storage pool and simply configuring them, instead of focusing on the physical storage capacity and storage media attributes of individual devices in the system, thus achieving unified storage management.
Compared with traditional storage, the advantages of virtualized storage are mainly reflected in: high disk utilization, disk utilization of traditional storage technologies is generally only 30-70%, and disk utilization after virtualization technology is as high as 70-90%; Flexible storage, which can adapt to different vendors and different types of heterogeneous storage platforms, providing better flexibility for storage resource management; convenient management, providing a centralized management method for large-capacity storage systems, and avoiding the expansion of storage devices. The management troubles are caused. With better performance, the virtualized storage system can perform load balancing. The bandwidth required for each data access is allocated to each storage module and the overall access bandwidth of the system is improved.
There are multiple classification methods for virtual storage, which can be divided into two major categories: classification based on the location of virtualization in the I/O path; classification based on different control paths and data paths. According to different locations where virtualization is implemented in the I/O path, virtualized storage can be divided into host virtual storage, network virtual storage, and storage device virtual storage. According to different control paths and data paths, virtualized storage is divided into symmetrical virtualization and asymmetric virtualization.
The system of virtualized storage constitutes a typical storage method in cloud storage as distributed storage. In this way, the storage device is generally managed in an out-of-band virtualization manner. Metadata management and data transmission are accomplished through an IP network. There are four main types of storage devices in this type of virtualized storage system.
Client:
Clients provide various application services to customers, such as web services, databases, file services, scientific computing, etc. The storage agent software runs on the client to provide read and write access to applications for network virtual appliance provisioning.
Configure the management server:
The configuration management server is used to configure and manage the system. Log in to the cloud storage platform through the Internet, Telnet, or other interface to remotely configure and manage the entire storage system.
Metadata server:
The cloud storage system's metadata server (MDS) manages the layout information of the entire system's metadata and object data, and is responsible for system resource allocation and network virtual disk address mapping. Globally virtualized storage management software and cluster management software deployed on the MDS can manage the configuration and operation of the entire storage system. In addition, the MDS implements redundant data relationships among common storage nodes through redundant management software.
Object storage node:
Each storage node is an independent storage device that is responsible for the storage, backup, migration, and recovery of object data, and is responsible for monitoring the operating status and resource status of other storage devices. At the same time, the storage node runs virtualized storage management software and stores the data needed by the application.
The three-tier model of virtualized storage According to the composition and characteristics of cloud storage systems, virtualized storage models can be divided into three layers: physical device virtualization layer, storage node virtualization layer, storage area network virtualization layer.
The three-tier virtualization storage model greatly reduces the complexity of storage management, effectively encapsulates the complexity and diversity of the underlying storage devices, and enables the system to have better scalability and flexibility. Users can create virtual volumes simply by adding the storage devices to the storage pool and simply configuring them, instead of focusing on the physical storage capacity and storage media attributes of individual devices in the system, thus achieving unified storage management.