What is virtualization?

Extract

[...] What is deduplication/Compression ? Compression refers to the process of reducing the size of a data. It is made with the repeat of pattern that can be made smaller. In order to do that, the use of algorithms is needed. Compression is used in the storage word for the backup and the archive. There are two kind of compression: - Inline compression: The data are compressed in cache memory before being drives on the back end. It needs a lot of volume and it is very slow. [...]

[...] We can see that object storage offer more possibility, more scalability and more performance than file storage. What is RAID? RAID means "redundant'array'of'inexpensive'disks". It was invented because of different factors. Before, the disk drives were expensive and they had limited IOPS. Usually, disk drives failed without warning. RAID was created to protect data from failed drives and provide higher performance of I/O. The following figures explained how RAID worked: As we can see, RAID combines multiple physical drives in order to create different logical drives. [...]

[...] It is usually written RAID 1+0. This level gives the best performance but it is an expensive system. The following figure shows how a RAID 10 is created: - RAID 5 is the most commonly use RAID level. In a technically way, RAID 5 is known as a block-level striping with distributed parity. It can work with different number of drives to 16). The figures shows a 3+1 RAID 5. The advantages are that the read data transactions are very fast and if one of the drive fails it will be possible to find the data in the other drives. [...]

[...] Another difference is that'object'is stored in with no hierarchy or tree-architecture as it is the case in a traditional'filesystem. This flat namespace is a key in the massive scalability inherent in object system storage. Nodes are the component of OSD system. A node is a server that allows the storage, the retrieving and the managing of the data in the system. The benefit of OSD is that, compare to SAN and NAS, security and reliability are provided. Also, it is an independence platform that combine scalability and manageability. We can use OSD in different cases. [...]

[...] The deduplication has to be done as the file level and not as the'block'level. As compression, there is two kinds of deduplication: - Inline'deduplication: data'are'duplicated on the'front'end before being committed to disk on the backend. - Post-Process Deduplication: without impacting the performance, post process deduplication involve the writing of the data on disk in a bloated form and check after if it is exist in another location. Compression provides an aggregate data reduction ratio range of 1.2:1 to 10:1 but it will have a nominal impact. [...]