RAID0 is actually non-redundant level and stands for striped. It joins together all the drives for maximum performance, but adding no redundancy what-so-ever. Full diskspace is accessible, but once a single drive fails the whole array fails. Expected MTTDL of 3.57 years.
RAID5 is a parity array, one of the drives is reserved for redundancy. Read speed is approximately 95%, but write speed only around 60%. On 4 disk array you will have 75% of space available, and on 6 drive array 83.33% of available space. Expected MTTDL of 2 531 years.
RAID5 is a parity array, one of the drives is reserved for redundancy. Read speed is approximately 95%, but write speed only around 60%. On 4 disk array you will have 75% of space available, and on 6 drive array 83.33% of available space. Expected MTTDL of 2 531 years.
RAID1 is a mirrored array, half of the drives is reserved for redundancy. Read speed can be upto roughly 190% depending on the use but typically at around 100% of one drive performance, while write speed is around 95%. You will only have about 50% of space available. Expected MTTDL of 12 480years.
RAID10 is a mirror + stripe (or stripe+mirror depending on the configuration). This is mostly same as RAID1 but much higher performance on average due to how the data is laid out to the disk, where as potential maximums remain the same in practice RAID1 often has the mirror drive doing nothing. You will have about 50% of space available. Expected MTTDL of 12 480years.
MTTDL stands for Mean Time To Data Loss. It is by no way a guarantee, and it only stands for given large enough pool it will average to this number. Take for example a RAID0 array of 3.57years, it is quite possible that some arrays fail within the first month, where as some array may last for 10 years. RAID5 for 4x8TB drives has very high MTTDL, but we have seen many RAID5 arrays fail during resync process. One reason is is because same model drive, from same factory, manufactured on the same date are much more likely to fail at the very same time. Even RAID1 and RAID10 arrays fail sometimes catastrophically.
We used the excellent calculator provided by Serve The Home at https://www.servethehome.com/raid-calculator/raid-reliability-calculator-simple-mttdl-model/ for these MTTDL numbers.
It is not question of IF but WHEN. All storage medias eventually fail, it is only question of time. This is why you should never keep anything you are not ready to lose on a RAID0 array, or at the very least you should have up-to-date backups constantly.
RAID IS NOT A BACKUP. So for important data you should keep backups no matter what level of redundancy.
RAID10 is a mirror + stripe (or stripe+mirror depending on the configuration). This is mostly same as RAID1 but much higher performance on average due to how the data is laid out to the disk, where as potential maximums remain the same in practice RAID1 often has the mirror drive doing nothing. You will have about 50% of space available. Expected MTTDL of 12 480years.
MTTDL stands for Mean Time To Data Loss. It is by no way a guarantee, and it only stands for given large enough pool it will average to this number. Take for example a RAID0 array of 3.57years, it is quite possible that some arrays fail within the first month, where as some array may last for 10 years. RAID5 for 4x8TB drives has very high MTTDL, but we have seen many RAID5 arrays fail during resync process. One reason is is because same model drive, from same factory, manufactured on the same date are much more likely to fail at the very same time. Even RAID1 and RAID10 arrays fail sometimes catastrophically.
We used the excellent calculator provided by Serve The Home at https://www.servethehome.com/raid-calculator/raid-reliability-calculator-simple-mttdl-model/ for these MTTDL numbers.
It is not question of IF but WHEN. All storage medias eventually fail, it is only question of time. This is why you should never keep anything you are not ready to lose on a RAID0 array, or at the very least you should have up-to-date backups constantly.
RAID IS NOT A BACKUP. So for important data you should keep backups no matter what level of redundancy.
