Why can C raid shorten the reconstruction time compared with traditional raid?

Fast reconstruction of CRAID (different from the traditional RAID method of kicking the disk first and then rebuilding, the fast reconstruction of CRAID can only rebuild the damaged data block on the wrong disk, directly copy the data block to the hot standby in the error-free area, and then transfer the wrong disk to IDDC disk diagnosis center for processing after reconstruction, which can obviously reduce the impact of the reconstruction process on the performance of RAID group.

In the traditional RAID reconstruction, the most time-consuming reason is that the unit disk needs to be called for XOR verification. In rapid reconstruction, the verification task is only a few percent or one thousandth of that of traditional RAID reconstruction, and the verification time can be ignored, while the speed of disk copying can reach the maximum value of disk reading and writing. Take 1TB SATA disk as an example. In) block RAID, the total reconstruction time is about 30 hours.

Fast local reconstruction

Rebuild only the data with differences, which is similar to rapid reconstruction, but it is not a hot standby. Only the changed parts of the original disk are rebuilt, so that we can live together. This method is suitable for the case that the disk is not damaged, but it is flashed or mishandled, and it is pulled out and inserted in a short time. This method can reconstruct the data lost when the disk is not available within 5 minutes, and the reconstruction time is short, which greatly reduces the damage degree of RAID group.

Rapid multiple reconstruction

Multiple failed disks can be rebuilt at the same time. If there are bad blocks in different units where two disks in a RAID group are located, both units can be rebuilt at the same time, thus improving the overall reconstruction.

CRAID greatly shortens the reconstruction time, greatly improves the redundancy, and can accommodate many bad disks without affecting the operation of the business.

Organize data with Cell as the core, break the binding relationship among LUN, RAID and disk, and form a complete virtualization architecture within the storage array.

Use the fixed granularity of 1GB to manage data nLUN, Cell and RAID, and there is no binding relationship between them. N spatial resources can be flexibly combined, and the cell can be obtained on demand and adjusted in real time according to the performance requirements.