1. Storage Level:
The access speed of the primary storage is much higher than the access speed of the disk. A classic data requires about 10 ms for disk access, while the memory only requires 60 ns !.
There is another important reason why data is stored in level-2 and level-3 storage. A 32-bit addressing system can only store 2 to the power of 32 bytes in the primary storage, the number of data objects may exceed this limit! In addition, the primary storage will lose data after the system is restarted (although it is possible to increase the storage power backup ). It is also important to optimize the storage of disk data records based on the disk's geometric and mechanical structures. If the two records are frequently used, they can be placed in the same block, gradually reducing the degree of proximity, in turn for a uniform track, the same cylindrical.
2. Cheap redundant disk array (RAID)
Disk is a potential bottleneck for system performance and storage reliability. The CPU increases by 50% every year, while the disk increases by 10%. In addition, the disk contains mechanical originals, so they have a much higher failure rate than the electronic components of computer systems.
Disk Arrays organize several disks together to improve performance and reliability. Data Division is used to improve performance, while redundancy is used to improve reliability. redundant information is not implemented by simply copying information, but is carefully organized to ensure disk failure, redundant information can be used to reconstruct the disk fault content. So far, in fact, this has become an industrial standard, and each raid level is actually a compromise between reliability and performance.