The so-called storage scheme is to manage the disk / disk group with separate software and hardware for the host.
Storage classification (Table 1 below) is divided into closed system storage and open system storage according to server type,
Closed system mainly refers to mainframe, AS400 and other servers,
Open system refers to the server based on windows, UNIX, Linux and other operating systems;
The storage of open system is divided into built-in storage and plug-in storage;
The plug-in storage of open system can be divided into direct attached storage (DAS) and fabric attached storage (FAS);
According to the transmission protocol, networked storage of open system is divided into network attached storage (NAS) and storage area network (SAN).
At present, most users use the open system, and the plug-in storage accounts for more than 70% of the disk storage market. Therefore, this paper mainly discusses the plug-in storage of the open system.
At present, there are three kinds of plug-in storage solutions
Direct attached storage (DAS)
Network attached storage (NAS)
San: storage area network
The hard disk used in the storage scheme is mostly SATA / SAS, and SSD is also used in pursuit of high performance. After serial connection / raid, it provides access interface to the host.
Table 2 is as follows:
Direct attached storage (DAS) refers to the direct connection of storage devices to a computer through SCSI interface or FC interface. DAS is not network storage, because only the host on which it is mounted can access it.
In other words, when the server fails, the data in the DAS storage device connected to the server cannot be accessed temporarily.
Das device example:
Das storage is widely used in the application of small and medium-sized enterprises. Because of its low cost, the storage system is directly connected to the application server. In small and medium-sized enterprises, many data applications must be installed on the directly connected Das storage.
Das storage relies more on the server host operating system for data IO reading, writing, storage and maintenance management. Data backup and recovery requires that the server host resources (including CPU, system IO, etc.) be occupied. Data flow needs to flow back to the host and then to the tape drive (Library) connected to the server. Data backup usually occupies 20-30% of the server host resources, Therefore, the daily data backup of many enterprise users is often carried out late at night or when the business system is not busy, so as not to affect the normal operation of the business system. The larger the amount of data stored directly, the longer the backup and recovery time, and the greater the dependence and impact on the server hardware.
The connection channel between direct attached storage and server host usually adopts SAS connection. With the increasing processing capacity of server CPU, the storage hard disk space and the number of array hard disks, SAS channel will become IO bottleneck; The sasid resource of the server host is limited, and the SAS channel connection that can be established is limited.
No matter the direct attached storage or the expansion of the server host, the expansion from one server to a cluster composed of multiple servers, or the expansion of the storage array capacity, will cause the downtime of the business system, which will bring economic losses to the enterprise, for the banking, telecommunications, media and other industries × 24-hour service of key business systems, which is not acceptable. Moreover, the upgrade and expansion of direct attached storage or server host can only be provided by the original equipment manufacturer, which is often limited by the original equipment manufacturer.
Storage area network (SAN),
There are two kinds of SAN storage at present
One is fcsan based on fibre channel;
The second is the Ethernet based ipscan (also known as iSCSI).
Fcsan is connected to the host (hbacard) through the optical switch, that is to say, the host that can be connected to the optical switch can access the storage;
As storage shared on Ethernet network, iSCSI is more similar to NAS.
Storage area network, from the name we can see that this is through the fibre channel switch to connect the storage array and server host, and finally become a dedicated storage network. After more than ten years of development, San has been quite mature and has become the de facto standard of the industry (but the optical fiber switching technologies of various manufacturers are not identical, and their servers and SAN storage have compatibility requirements).
San provides a simple way to connect with the existing LAN, and supports the widely used SCSI and IP protocols through the same physical channel. San is not limited by the current mainstream layout based on SCSI storage structure. Especially important, with the explosive growth of storage capacity, San allows enterprises to increase their storage capacity independently. The structure of San allows any server to connect to any storage array, so that no matter where the data is placed, the server can directly access the required data. Because of the optical fiber interface, San also has higher bandwidth.
Because San solution is to separate the storage function from the basic function, it is unnecessary to consider the impact of backup operation on the overall network performance. San also simplifies management and centralized control, especially when all storage devices are clustered together. Finally, the optical fiber interface provides a connection length of 10 km, which makes it very easy to achieve physically separate storage out of the computer room. San is the mainstream standard of high-end storage in the industry.
Fcsan device example:
IPSAN device example:
Network attached storage (NAS) refers to connecting storage devices to a group of computers through a standard network topology (such as Ethernet). NAS has a file system and IP address, which can be similarly understood as the shared disk of network neighbors.
NAS is a file level storage method, which focuses on helping working groups and department level organizations to solve the demand of rapidly increasing storage capacity. Nowadays, users use NAS for document sharing, picture sharing, movie sharing and so on. With the development of cloud computing, some NAS manufacturers have launched cloud storage function, which greatly facilitates the use of enterprises and individual users.
NAS products are really plug and play products. NAS devices generally support multi computer platforms, users can access the same documents through network support protocol, so NAS devices can be used in mixed UNIX / windows LAN without modification, and the application of NAS is very flexible.
But NAS has a key problem, that is, the bandwidth consumption in the backup process. Unlike storage area network (SAN), which transfers backup data stream from LAN, NAS still uses network for backup and recovery. One disadvantage of NAS is that it transfers storage transactions from parallel SCSI connections to the network. That is to say, LAN must handle not only the normal end-user transport stream, but also the storage disk request including backup operation. NAS performance is much worse than San, but the relative cost is also much lower.
NAS device example:
Editor in charge: CC