Memory for disk arrays
It should be noted that with parallel access model is implemented only on the
condition that the size of the request for recording the data of the size of the
block itself. Otherwise realize Pas dual listing several blocks, it is simply
impossible. Imagine the situation where the size of a separate unit is 8
kilobytes, and the size of the request for data entry - 64 Kbytes. In this case
study information on Debian GNU cut into eight blocks to 8 Kbytes each. If there
is an array of four discs, it can simultaneously record four blocks, or 32
Kbytes, for once. Obviously, in the example considered by the burning speed and
the speed of reading would be four times higher than when using a single disk.
However, this situation is ideal, because it is not always the size of the
request, and a multiple of the block size, the number of disks in the array. If
the same amount of recorded data block size smaller, it is fundamentally
different access model - independent access. Furthermore, this model can be
implemented in the case where the amount of recorded data of the size of one
block. When independent access all the data are recorded on a separate request
for a separate disc, the situation is identical with one disk. The advantage of
the model with independent access that while the flow of several requests for
the record (reading), they all will be carried out independently on separate
disk Such a situation is typical, for example, in servers.
Next part
|
O
wlCom Software
