Chinese
RAID1- - 5 technologies details is solved (below)
From;    Author:Stand originally
RAID 3

What RAID 3 uses is way of a kind of relatively simple desired result implementation, use a special disk to deposit all desired result data, and the market that head a division is established in the rest disk of dispersive data read keep an operation. For example, in a RAID 3 system that makes by 4 hard disk, 3 hard disk will be used to save data, the 4th hard disk is used technically at desired result. This kind of form that configures means to be able to use 3 1 expresses, specific if pursue:

The picture is browsed in new window by this


Going up in the graph, we express to use with same color same all data of a desired result piece piece, the part of diagonal line out is desired result piece. Desired result piece and the data of all correspondence piece form division of a belt together.

What the place of each desired result piece in the 4th hard disk includes is other data of the correspondence in 3 hard disk piece desired result information. The successful place of RAID 3 is providing fault tolerance function in that way at can resembling RAID 1 not only, and whole pays expenses from RAID 1 50% drop for 25% (RAID 3 1) . Use the grow in quantity of disk amount as place, cost expense will be less and less. E.g. , if we use 7 hard disk, so total expenditure arrives with respect to meeting general 12.5% (1/7) .

Below different situation, RAID 3 reads the complex rate that keeps an operation to differ. The simplest case reads access to occupy from inside a whole RAID 3 system namely. At this moment, need to store in data only dish in find relevant data piece undertake reading extraction operating can, won't increase any additional system expense.

When keeping data to RAID 3, the situation can become a few more complex. Although we just write a data piece to a disk, also must calculate what with this data piece manages division of a belt together is all the desired result value of data piece, in keeping new cost desired result piece afresh. For example, when data piece writes the green in pursueing up when us into data, must recount the desired result of data piece is worth 3 all green, rescript next the green desired result that is located in the 4th hard disk piece. From this we can see, one wrote an operation to include data to read in fact take (read take the related data in heading a division piece) , desired result is worth computation, data piece is written and desired result piece keeps 4 processes. Systematic expense increases greatly.

We can take the size of the area to make RAID system gets simplifying through proper setting. If a certain length that keeps an operation is equal to the size of area of a whole area fitly (whole area area is written) , so we need not reread takes the correlation in heading a division value of desired result of data piece computation. We need to calculate the desired result of whole area area to be worth only, keep data and desired result information into data directly next dish with desired result dish can.
Previous12 Next