There is a story circulating in the field of virtualization: A good virtualization solution is like a theme park that travels through a virtual reality. When a visitor imagines that he is gliding over the city, the sensor will convey the corresponding real feeling to the visitor while hiding the real mechanical environment. Similarly, a good virtualization tool can do the same for an enterprise storage device, but the process may be reversed—first create a framework that lets the data feel stored in a real physical environment, after which the operator It is possible to arbitrarily change the location of the data storage while ensuring centralized security of the data.
Host-based virtual storage relies on agents or management software that are installed on one or more hosts to enable storage virtualization control and management. Since the control software is running on the host, this will take up the processing time of the host. Therefore, the scalability of this method is poor, and the performance of actual operation is not very good. Host-based methods may also affect the stability and security of the system, as it may result in inadvertent access to protected data. This approach requires the installation of appropriate control software on the host, so a single host failure can affect the integrity of the data throughout the SAN system. Software-controlled storage virtualization can also introduce unnecessary interoperability overhead due to differences in hardware and software between different storage vendors, so this approach is less flexible.
However, because no additional hardware is required, host-based virtualization is the easiest to implement and has the lowest equipment cost. Vendors using this approach tend to become software vendors in the storage management arena, and there are already mature software products available. These software provide an easy-to-use graphical interface for easy SAN management and virtualization, with a good load balancing mechanism in both host and small SAN fabrics. In this sense, host-based storage virtualization is a good value for money.
Method 2: Virtualization based on storage devicesStorage-based storage virtualization methods rely on storage modules that provide related functionality. Without third-party virtualization software, storage-based virtualization often provides only an incomplete storage virtualization solution. For SAN storage systems with multi-vendor storage, this approach does not work very well. Functional modules that rely on storage vendors will exclude JBODS (JustaBunchofDisks, simple hard disk groups) and simple storage devices from the system, as these devices do not provide storage virtualization capabilities. Of course, using this approach means eventually locking up a single storage vendor.
The storage-based virtualization approach also has some advantages: it is easier to implement in a storage system, easier to coordinate with a particular storage vendor's device, so it's easier to manage, and it's transparent to users or managers. . However, we must note that because of the lack of sufficient software to support, this makes the solution more difficult to customize and monitor.
Method 3: Network-based virtual storageThe network-based virtualization method implements storage virtualization between network devices, in the following ways:
1. Virtualization based on connected devicesIf the method based on the interconnected device is symmetrical, then the control information and the data are on the same channel; if it is asymmetric, the control information and the data are on different paths. In a symmetrical manner, interconnected devices can become a bottleneck, but multiple device management and load balancing mechanisms can alleviate bottlenecks. At the same time, in a multiple device management environment, when a device fails, it is easier to support the server to achieve fault replacement. However, this will result in multiple SAN islands because one device only controls the storage system to which it is connected. Asymmetric virtual storage is more scalable than symmetric because the path of data and control information is separate.
The interconnected device-based virtualization approach can run on a dedicated server using a standard operating system such as Windows, SunSolaris, Linux, or a vendor-supplied operating system. This method runs on a standard operating system and has many advantages based on the host method—easy to use and cheap. Many device-based virtualization providers also offer additional functional modules to improve overall system performance, achieve better performance and better functionality than standard operating systems, but require higher hardware costs.
However, the device-based approach also inherits some of the pitfalls of host-based virtualization because it still requires a proxy or host-based adapter running on the host, and any host failure or inappropriate host configuration can result in access. To data that is not protected. At the same time, interoperability between heterogeneous operating systems remains a problem.
2. Router-based virtualizationThe router-based approach is to implement storage virtualization on the router firmware. Vendors typically also provide additional software running on the host to further enhance storage management capabilities. In this method, a router is placed in the data channel of each host to the storage network to intercept any command from the host to the storage system in the network. Since routers potentially serve each host, most control modules exist in the router's firmware, which is better and better than the host-based and most interconnected device-based methods. This approach is more secure than a host-based or device-based approach because it does not rely on a proxy server running on each host. When the router connecting the host to the storage network fails, it may still cause the data on the host to be unreachable. However, only hosts that are connected to the failed router will be affected, and other hosts can still access the storage system through other routers. Router redundancy can support dynamic multipathing, which also provides a solution to the above problem. Because routers often serve as a bridge for protocol conversion, router-based approaches can also provide interoperability between heterogeneous operating systems and multi-vendor storage environments. [2]
80 Plus Gold
Boluo Xurong Electronics Co., Ltd. , https://www.greenleaf-pc.com