1. Summary of cloud storage
Cloud storage is a kind of online storage (English: Cloud storage) model, namely the data stored in usually by a third party hosting more than one virtual server, rather than a dedicated server, hosting, hosting companies operating large data centers, need data storage custody, through to its purchase or lease storage space, to meet the needs of data storage, data center operators according to customer demand, the back-end to storage virtualization of resources, and to store the resource pool (storage pool) way to provide the customer can use this on its own storage resource pool to store files or objects, in fact, these resources could be distributed in many server host,
Cloud storage service is through a Web service application program interface (API), or through the Web user interface to access,
Storage:
Storage technology is one of the core technology of computer, computer storage technology (such as hard disk, network storage, virtualization storage technology) storage capacity and I/O is the general trend of increasing speed, with the development of information technology, emerging new storage technology, storage industry important history as follows:
The first hard disk storage,
The world's first hard-disk storage IBM 350 ramac was born, was it the total capacity of only 5 MB, but with a total of about 50 24 inches diameter disk,
RAID technology,
SAN technology,

2. The classification of storage
Depending on the type of server storage can be divided into a closed system of storage and storage of open system, open system of storage is divided into internal and external storage, among them, external storage can be divided into direct storage (DAS) and network storage (FAS), according to different forms of network, the current three mainstream storage or storage solution for direct type storage (DAS), storage area network (SAN), the network access storage (NAS),
In the face of DAS, NAS, SAN comparison:
Direct connection refers to the storage device by the SCSI interface directly connecting to a server, storage medium directly articulated on the internal bus, the data storage is a part of the entire server structure,
The advantages of DAS
Simple installation, storage equipment cheap,
The disadvantage of the DAS
Don't have the ability to disaster preparedness and only suitable for small-scale (household, small businesses use),
Not suitable for the performance requirements of high scenarios,
NAS devices essence is to file system of the local host IP network devices, multiple user nodes can share the same file system on a NAS,
NAS is a thin server on the network, the user's ability to provide a file system access, management, storage medium and in charge of data backup, we usually open the samba on x86 server services, and then on Windows mapping a drive letter to the samba server can use like to use local storage on the server storage, at this moment, this server can also be seen as a NAS server,
The FTP server calculating do not calculate the NAS? Not, because the customer access to the file on the FTP server, also need to download the complete file to the local file system directories, and on the NAS network file system is can not rely on the local file system, mount directly use (such as the current commonly used NFS, CIFS file system),
SAN is through a dedicated high-speed network to one or more network storage devices and servers connected dedicated storage system, can interpret SAN into a network, the network contains a variety of elements, such as disk arrays, FC switches,
In the SAN
SAN can be divided into two types of FC_SAN and IP_SAN,
FC_SAN
SAN only FC - SAN, a general says SAN default means FC_SAN, it USES the fiber channel to transmit data, it USES is the SCSI protocol,
IP_SAN
IP_SAN was produced after FC_SAN, it USES through IP network data transmission, using iSCSI transmission standard SCSI protocol data, comparison with IP_SAN FC_SAN
FC_SAN advantage:
Transmission bandwidth is high, there are 1, 4-trichlorobenzene and 8 gb/s four standard, the mainstream is 4 and 8 gb/s
Performance is stable and reliable, mature technology, is the key application fields and mass storage network choice,
FC_SAN faults:
Cost is very high, need optical switches and a large amount of fiber optic cabling;
Maintenance and complex configuration, need training is completely different from LAN administrator professional FC network administrators,
IP_SAN advantage:
Cheap, buy cables and switches are using Ethernet, even can make use of existing network to form a SAN.
Easy deployment, low management difficulty;
Wan Zhao Ethernet the advent of IP SAN when compete with FC SAN is no longer as transmission bandwidth;
Based on IP network is a natural advantage makes IP SAN storage is easy to implement different ground, disaster remote to time through WAN technologies, such as
IP_SAN faults:
IP_SAN in many other aspects of performance FC_SAN or discriminating, proprietary spheres FC_SAN,
DAS, operating system, file system, storage is a whole,
NAS, the file system and storage for separating out,
In SAN, only give independence, storage file system is still in the access side,

3. The RAID
Raid 0 mode=total capacity disk number * disk capacity
Advantages: in the condition of raid 0, storing data is divided into two parts, respectively stored in two hard disk, mobile hard disk at this time of the theory of single disk storage speed is 2 times, the actual capacity is equal to two pieces of hard disk in a minimum of 2 times,
Faults: any piece of hard disk failure, the data on the whole raid will not resume, note: high-definition movie is suitable for storage. The
Raid 1 mode=total capacity (number of disk/2) * disk capacity
Advantage: this mode, the two pieces of hard disk mirror image, when a hard disk is damaged, replace a new hard drive (greater than or equal to the original hard disk capacity) to replace the original hard disk can automatically restore data and continue to use, mobile hard disk of the actual capacity is equal to a smaller disk capacity and speed is the same as single disk storage. The raid 1 has the advantage of any one piece of hard disk failure, the stored data is not lost,
Weakness: the model can be used hard disk actual capacity is small, the smallest hard disk for only two hard disk capacity,
Note: very important information, such as databases, personal data, it is quite safe storage solution
Raid 3 total capacity=(disk number - 1) * disk capacity
It with 2 is very similar to RAID, are the data market tumble throughout different hard disk, the difference is that RAID 3 USES simple parity, with single piece of disk storage parity information, if a disk failure, odd-even plate and other data plate can produce data; If the failure does not affect the data using odd-even plate, RAID 3 for a large number of continuous data can provide a good rate, but for random data, odd-even plate will become the bottleneck of write operation,
Raid 5 mode=total capacity (disk number - 1) * disk capacity
Raid5 wrong data storage for backup, but the data and the parity information stored in the corresponding raid5 each disk, and parity information and the corresponding data are stored in different disk, a disk data when raid5 after damage, using the rest of the data and the corresponding parity information to restore the damaged data,
Raid 6 total capacity=(disk number - 2) * disk capacity
RAID 6 compared with RAID 5, RAID 6 adds a second independent parity information block, two independent parity system using different algorithms, the reliability of data is very high, even if the two disk failure will not affect the use of the data, but the RAID 6 need to be assigned to parity information more disk space, relative to the RAID 5 have greater loss "write", so "write performance is very poor, poor performance and the implementation of the complex in such a way that a RAID 6 is rarely practical application,
Raid 10=total capacity (number of disk/2) * disk capacity
Raid10 need at least disk to complete, the two pieces of hard disk of a raid 1, then two sets up a raid 0, raid 1 although raid10 scheme caused 50% of the waste of disk, but it provides 200% of the speed and damage of single disk data security,
Raid the total capacity of 50=(disk number - 1) * disk capacity
RAID50 is the combination of RAID5 and raid 0, this configuration in RAID5 son the disk groups, including parity information on each disk data, each RAID5 child requires three hard disk group, RAID50 higher fault tolerance ability, because it allows a group with a disk failure, and not cause data loss, and because the parity bit division in RAID5 disk group, so it has a lot to improve the reconstruction speed, advantages: higher fault tolerance ability, have read faster data rate potential, it is important to note: disk failure will affect throughput, rebuilt after fault information time is longer than mirror configuration cases,

4. Disk hot standby principle
The hot spare disk equivalent to do multiple backup for Raid arrays more, if the Raid on display in one of the disk is broken, the hot standby plate will replace the bad disk in the Raid, at the same time the use of different or calibration algorithm, the broken plate of the above data is done and stored in the hot plate, so that is equal to Raid damage, and then you find one the same disk replaced the bad disk, and thermal for Raid and the state of normal again!

5. The concept of data storage of
Data storage is the data stored in different types of storage devices, through hierarchical storage management software to realize the automatic migration between entity data into a storage device; According to the data access frequency, retention time, capacity, and other factors, such as performance requirements determine the optimal storage strategy, so as to control rules of data migration, the advantages of hierarchical storage:
1) the maximum meet user requirements,
2) reduce overall storage costs,
3) performance optimization,
4) improve data availability,
5) data migration transparent to application
Data storage (on - line) storage generally fall into line, near line (near - line) and offline (off - line) storage three storage mode,

6. The concept of cloud storage and technology principle
Is generally recognized by the industry of cloud storage two mainstream technology solutions: distributed storage and storage virtualization,
Distributed storage mainly includes distributed storage, distributed file storage, distributed object storage and distributed storage four types, table
Distributed storage, is simply provides a piece of equipment storage interface, users need to block storage volume attached to the virtual machine (or equivalent) before can interact, these volumes are persistent, they can be lifted from the running instance or reattach the data remain intact,
Features: simple,
Application scenarios: used to increase the I/O skills
Distributed file storage, designed for enterprise applications to customize distributed file storage system, adopts the distributed architecture, no single point of the whole system, no performance bottlenecks, flexible support elastic expansion, support directory level quota management, intelligent operational interface,
For all types of enterprise applications to provide alternatives to traditional NAS, FTP, a new generation POSIX file storage solutions, for enterprise data storage performance effectively, improve the availability, reduce operational costs,
Distributed object storage, adopting the distributed cluster architecture implementation, for the enterprise massive amounts of unstructured data to provide data storage solutions, provide Amazon S3, it Swift standard object storage agreement, standard HTTP load without a single point, without the bottleneck of high performance, flexible expansion, the object of high availability storage solution, which can be widely used in the enterprise network backup, image data, Internet applications, the data backup all sorts of unstructured data storage, such as object maintains its own properties, simplifies the storage system management tasks, increased flexibility,
Part of the object storage system
The basic unit of the 1) objects: object storage,
2) object storage devices: the core of the object storage system,
3) file system: to explain the user file operations, and in the metadata server and the communication between object storage devices, complete the requested operation,
4) the metadata server: provide the client with metadata,
5) network connection: object storage system important constituent,
Distributed table stores, is ali cloud itself no more model database, provides mass structured data storage and fast query and analysis services, table storage index of distributed storage and powerful engine that can support petabytes of storage, thousands of TPS and millisecond delay service ability,

7. The object storage technology
nullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnullnull