User demand for storage is not only in terms of volume, but also data transmission performance, data management, storage scalability, heterogeneous product integration and other aspects.To meet these requirements, storage virtualization gradually mature and become an irreversible trend.
For many businesses, the storage system is the IT system, a completely separate level, like the network, applications and servers, and storage system in the status of IT systems has become increasingly important.
Enhance the status of the storage system, directly stimulate the growth of its market.According to the relevant institutions predict that from 2004 to 2009, China external disk storage systems for the growth rate of 13.1% growth in overall storage hardware market, compared with 10.2%.
To snatch the storage of this huge market, some companies issued "as the storage of Chinese cabbage to sell" argument, but to a dominant position in the competition, the manufacturers are also vigorously promote the development of storage technology.
In this environment, the virtual storage technology to mature by the obscure corners come to the fore.
What is virtual storage
Virtual memory is the storage virtualization system.Typically, virtualization is the virtual abstraction layer to achieve.The physical storage system will store the body of logically organized to form a logical memory banks (also known as the storage pool), then depending on the needs of the storage pool is divided into virtual disk or virtual volumes.Use of virtual volumes, making the application and presented to the user is a physical disk abstraction.So that each server or workstation can see the logic in the local storage unit and the local hard drive does not make any difference.
In a virtual storage environment, the capacity of a single storage device, speed and other physical features are to be blocked, no matter what the background of the physical storage devices, servers and applications that customers are very familiar with the logic of the image storage deviceTherefore, system administrators do not care about their back-end storage, just focus on management of storage space itself.All storage management operations, such as system upgrades, change the RAID level, logical initialization, create and allocate virtual disk storage space geometric expansion of any previous storage technology is more easily achieved.
Virtual storage is how to produce
Can say that the technology of virtual storage is not new.Virtual storage is developed together with the computer technology, and was first started in the 70s of last century.At that time, storage capacity, especially in the memory capacity is small, the cost is very high, a large application or program application has been limited.To overcome the limitations of virtual storage technology to produce, the most typical application is the virtual memory technology.
VMS uses the early time-sharing system, also used a mainframe virtualization technology into multiple virtual computers, will be two or more computers into a computing system virtual entity.
Widespread use of RAID technology is actually a multiple physical storage media (disks, tapes) into a large virtual logical storage medium.
Therefore, the concept of storage virtualization in the computer field has actually been widely used.
Explosive growth of data storage equipment, networking and diversity, computer technology and related information processing technology continues to evolve, making it the growing demand for storage.This stimulated the emergence of new technologies, such as getting better and better disk performance and capacity is growing.However, in a large number of medium-sized information processing system, a single disk can not meet the needs of this case storage virtualization technology to developing.
In this process there are also several stages and several applications.First, the disk stripe set (RAID, fault tolerance can be with) technology, multiple physical disk through a certain logic together into a large virtual disk capacity.
The increasing amount of data required and the continuous improvement of data availability, but also a new storage technology came into being, that is, storage area networks (SAN) technology.
Wide area of SAN storage devices are designed to achieve as a public utility, any officer, any server can access their desired data anytime, anywhere.
Currently under discussion, including more iSCSI, FC Over IP technology, although some related virtual storage standards have not yet finalized, but wide area of storage networking, storage virtualization is an irreversible trend.
What are the advantages of virtual storage
Expand more easily: virtual storage devices to avoid the expansion of the storage management brought trouble.
For example, the use of general storage system, when adding new storage device, the entire system configuration needs to be tedious work, we can make this "new members" added to the storage systems.The use of virtual storage technology, adding new storage devices, network administrators need only simple storage system, the system configuration changes, the client does not require any action that feels just increased the storage capacity of the system.
Improve the overall bandwidth: a virtual memory storage system can greatly improve the overall access bandwidth.Storage system is composed of multiple memory modules, the virtual storage system can be a good load balancing, data access to every reasonable distribution of the bandwidth required to each memory module, so that the system increases the overall access bandwidth.
For example, a storage system with 4 memory modules, each memory module access bandwidth of 50MBps, then the storage system can approach the total access bandwidth to the bandwidth of the memory modules and, that is 200MBps.
Integration of heterogeneous products: Virtual storage for storage resource management provides better flexibility, can be different types of storage devices to use centralized management to protect the user previously purchased storage equipment investment.
Improve storage efficiency: virtual storage technology to solve the waste of storage space used, it dispersed the system memory space of each integrate to form a continuous addressing of the logical storage space, breaking the single physical disk's capacity limits, and because the storage poolextensions can automatically redistribute data and the use of efficient snapshot technology to reduce capacity requirements, thereby greatly enhancing the utilization of storage resources.
Support for business continuity: If deployed properly, the virtual store can give administrators greater flexibility to modify the infrastructure without affecting the system or application.Virtual memory by increasing the flexibility of the infrastructure, making more independent management of IT infrastructure instead of relying on applications and business activities, simplifying set of management.
For example, each have their own storage arrays and server configuration volume methods, in the future, the virtual store will also provide a separate method of storage administrators, uniform standards within the network to achieve these functions.
Other features: virtual storage management software can, for the network system provides a number of other useful features, such as no server remote mirroring, data snapshot (Snapshot) and so on.
How virtual storage
Networked storage is the inevitable development of future storage, both in the NAS in the network environment or in the SAN network environment, an essential part of their network can be summarized into three levels: for client server systems, providing storage resourcesstorage device, attached storage and server core switching equipment.Therefore, the virtual store can also be divided into: server-based virtualization, network-based virtualization, storage-based virtualization.
Server-based storage virtualization: the virtual server-based storage virtualization layer through the implementation on the server.This implementation does not require additional special hardware, the virtualization layer in the form of a software module embedded into the server's operating system, the virtual layer driver module as an extension for a variety of storage devices connect to the server to provide the necessary control functions.
For example, it can be connected to a group level in an array of small-scale spatial configuration, forming a large space; also a large array can be divided into manageable volume of small units; or space in the maintenance of an application canto extract a number of spatial data in the array.
Because does not require any additional hardware, server-based virtual storage the most easy to implement, the lowest cost.There are already such a mature software product.The software provides easy to use graphical interface for storage management and virtualization, the server and a small storage system has a good load-balancing mechanism.
Web-based data channel mode: the data structure is a channel between the server and the array of network data channel into the virtual storage management device.These devices typically provide space management and other supporting functions, such as data migration and copy services.
Data channel structure using large-capacity cache, significantly improve data transmission speed.Caching is widely used in storage servers and storage devices located between the I / O path in the middle of the media.
When the server reads data from the storage device, the physical location associated with the current data is stored in the cache of data read, and the frequent calls in the cache data retention; when the server is reading data, the probability that from a largefind the required cache data read directly from the cache.
Outside the data channel based approach: the design of passage, by way of servers and storage devices in the data transmission channel between the outside, by configuring a virtual memory manager to implement virtualized storage pool.Virtual Storage Manager via its FC ports connected to storage networks and provides a central management point, the entire storage network for centralized management.
Passage, by the way design is separated from the data stream management information, to avoid the channel structure of the internal operational issues arising.Virtual storage controllers in the array of different physical hard disk capacity of the logical combination of the band of virtual sets, multiple array controller port binding, to a certain extent, improve the system's available bandwidth.
Virtual storage controllers is not the actual data path, it will not be a hardware performance bottlenecks in the system bandwidth at the same time, even if it fails, it will not cause obstruction of the data network system to improve the security of the system.
Controller-based virtualization: controller-based virtualization, also known as "box" virtualization, it is the virtualization layer on the storage device adapters, controllers, etc. up to achieve.
The way to achieve optimal performance in theory, it can fully consider the physical characteristics of storage devices and server implementation of the work from the virtual storage freed directly implemented on the storage device, the method is simple.
In a storage device, the box solution has the advantage of easy to configure, and advanced programs can also control the communication between the multiple storage devices to the common virtualization.
Box virtualization can greatly reduce the complexity of storage deployment, enable administrators to the storage resources as a combined resource pool management, capacity management and the independent units from the original level of increase to the pool level.
Box virtualization architecture also eliminates other common so-called "standard capacity."Standard size is usually considered pre-assigned to a volume capacity and the capacity for future growth in demand for ready.But the box virtualization eliminated the traditional demand that the volume expansion of the work has become easier, and provides higher capacity than traditional rate structure.
There is also a controller-based virtualization solutions, is to be called the Network Storage Controller IDC storage device, which is also based on the controller, in essence, virtualization, but virtualization inside the box with the normal of different network storage controllercan also be connected to the top of it to virtualize other storage device, so the network storage controller, also known as the external controller-based virtualization platform.
The challenge of virtual storage
Can see that the function of a variety of virtual storage is very impressive.However, a variety of storage solutions have their limitations.
Server-based virtual storage system, because it is working on the server, it must be set separately for each server configuration and application.If the server is a small number, this is not a problem.However, the system in an enterprise setting, usually have hundreds or even thousands of servers to store data through the SAN system, control the distribution of space soon became a serious problem.
Second, the virtualization layer that resides on the server, which can be embedded in software modules to be into all types of operating systems, increasing the difficulty of software.
Again, the need for a centralized management strategy, this virtualization technology is actually implemented in a distributed environment, when a server of any malicious or illegal data operations, it may affect all connected to the storageequipment, data integrity and consistency, requiring an appropriate focus on management strategies.
The main drawback in the approach channel is a path in the network adds an additional "jump", which increases the server and the time delay between the physical storage devices.
Some channels in the device inside the device tries to use the cache in the solution to the problem of waiting time.But this largely depends on the application and caching algorithms.
Since the control of virtual storage directly to existing switching equipment and storage devices on the server, all servers have access to the storage devices and management through its channels.In order to control the switching equipment does not become a bottleneck of the entire system bandwidth, the device requires a large-capacity cache to be used for data exchange.
The same control as the exchange of virtual storage devices directly to exist on the server and storage devices, so the safety of the device is critical for the entire system, once it fails, all data channels will be blocked, resulting in data transmission network systemcollapse.
Virtual channel structure is a more serious drawback limits its scalability.Since all regions in the virtual I / O channel in the way of need, not in terms of bandwidth, processing power is, will form a bottleneck.Once any kind of resource depletion, we must take the data expansion strategy.This requires increasing access nodes to achieve passage of large-scale virtualization.In this way, through virtualization reduce the complexity and simplify management of the mind can not be achieved.
Corridor means the security of virtual storage is relatively poor.It is, in essence, or stripe set - disk array structure, once the band focused on a disk array controller is damaged, or the array of copper on the path to the switch, GBIC damage, will lead to a virtual LUN offline, andstripe set itself is not fault-tolerant, a LUN of the damage means the loss of data inside the Strip.