EVGA Z370 FTW (134-KS-E377)
- 56 -
Similar to RAID1, or any other current type of array with fault tolerance, a RAID5 array
is still usable even while it is experiencing a missing or failed drive resulting in the array
functioning in a degraded state. Performance will suffer in a degraded state until the
missing drive is replaced and the software rebuild process is completed. During the
rebuild process, performance will be severely degraded and can take several hours,
depending on the size of the array and percentage of capacity used.
The Good-
Most space efficient array-type that also has fault tolerance.
Initial array builds quickly
Read/write speed is very good; faster than RAID1 but slower than RAID0
The Bad-
Rebuilding an array can take a long time on an integrated PCH controller,
especially when using a very large array with multiple drives or capacity.
o
For this reason, RAID5 is sometimes more beneficial on a dedicated
RAID controller.
If more than one drive fails at once (not likely, but still possible) you will suffer
a total loss of data; the array will behave like a RAID0 losing a drive.
RAID5 is close to outliving its usefulness because it was created, in part, to
create large size arrays with fault tolerance. Due to the capacity available on
modern drives, other RAID solutions are now better and with fewer downsides.
RAID5 protects data rotating parity (there are several terms coined by different RAID
manufacturers over the years for RAID5 and they all mean roughly the same thing),
which means taking small portions of data, duplicating them and putting them onto
different drives. When a drive that has failed is replaced, its data is recreated by the
remaining drives in the array, which will require a very high volume of small data
segments copied back over to the replacement drive while also maintaining the array’s
index of data. Once the rebuild is complete, current data can be read from the new
drive and new data can be written to it.
The RAID5 diagram below is a bit more complicated than the others, due to the nature
of how data is distributed. In the diagram you will see that there is one more Physical
drive than there is data set. This method of data distribution shows that as long as you
have three (3) copies of each data set (Data-A, Data-B, and DATA-C), the array will be
functional and capable of rebuilding when you add in replacement drive in. And while
this is not mathematically correct for HOW the data distribution works, it is a good
visualization to understand the basics of how it works.
