Is it worth using hardware RAID controllers with a ZFS-based Open-E JovianDSS?
Our support experts often hear the question if it’s ok to use RAID controllers with…Read More
Nowadays, constant data access is a basis of many industries to ensure fluid operations. Lack of such access can result in a loss of credibility among customers, a complete halt to the operation, or even bankruptcy. Therefore, choosing a method for data management by selecting the proper storage setup is crucial. So the question arises – which factors should you pay attention to once deciding on a particular RAID? Here is some advice from us you can consider regarding the method of disk space management in your array.
The first factor we take into account is data read and write speed which is crucial in such industries like gaming or computational science. The best solution here would be implementing RAID 0 to the array. This method combines the capacity of two disks to create a single, vast, and accessible space that provides a logical drive that contains all the data. It can be achieved by adding the equivalent capacity of the smallest disk mounted. High read and write speed is possible because the processes happen parallelly on both disks in the array. As the data is divided into smaller units the process can be run on both disks simultaneously.
However, there are some downsides. The main one is that RAID 0 sometimes is not even called a RAID. The reason is the safety level – the lowest of all available options. There is no redundancy, no parity information, and no fault tolerance. Thus, if you want to give up at least a bit of write speed, using RAID 1 in the array seems a fair solution. The data will be mirrored, so only be copied onto the second drive in case of one disk fault. The write speed is fair, and the read speed is still remarkably fast. Paraphrasing a quote taken from the Iron Maiden song…
“…a single par(i)ty call on me (…) at the speed of
But what in case there’s a need for the data to be redundant, you may ask?
There are many possibilities, but you should be ready for further compromises. RAID 5 and RAID 6 allow you to implement redundancy since files are spread striped across all disks differently. RAID 6 is a better but more expensive solution. More disks are involved, and the parity happens on one more drive than with a RAID 5 solution. Ok, so how exactly is it possible, and when does it come in handy? Well, in the case of data safety.
RAID 5 is the best if your business relies on this factor. This RAID uses at least three drives in the array. The data is spread between them by dividing it into two or more parts (e.g., A1 and A2) and adding parity data (e.g., Ap), which is placed on the last drive. That allows for the failure of an entire single disk without risking any data loss. RAID 6, on the other hand, is based on the RAID 5 functioning (creating more chunks and striping them along the available storage, e.g., A1, A2, and A3), but thanks to adding another disk to the array structure, the parity data is doubled (e.g., Ap and Ap’). We can afford to lose up to two disks without losing any data. As you can see, it is more reliable than RAID 5, but the costs are higher because more drives are needed. You can also enhance both RAIDs with one or two spare drives (it will be respectively 5/6E or 5/6EE RAID). However, this is a far more expensive solution as there are more disks to be replaced once they fail. Another disadvantage is reducing the read and write speed, as the system uses more resources to calculate the checksums and perform a rebuild after the failure.
Another significant factor in data storage is… the amount of storage itself.
Once the data is on our drives in the array, we need to make sure how much of it we can keep on our disks and how much capacity can be used for mirroring or parity. RAID 0 does not offer any of that, so all we have is storage. RAID 1 gives more than 50% capacity for the stored data, and the other 50% remains for mirroring. RAID 5 and 6 are the best scorers here. Exactly one parity data chunk (e.g., Ap, Bp, Cp) is on each drive in RAID 5. Therefore, it allows for one drive failure. RAID 6, on the other hand, doubles the amount of parity data (two chunks, e.g., Ap and Ap’, Bp and Bp’, Cp and Cp’, etc.) allowing for two disk failures. Comparatively, it is much safer than RAID 5.
You may also consider choosing a software or hardware solution to control your RAIDs. Certainly, there are advantages and disadvantages to both. Hardware RAID solution allows you to increase the read/write performance as the processor or memory usage by the server controller is reduced. In addition, it can even improve the efficiency of the systems with low processing power equipment. However, it requires additional expenses related to purchasing, replacing, and fixing issues caused by using hardware. Software RAID solution, on the other hand, provides you with higher storage efficiency, especially if you consider using Open-E JovianDSS. Setting up a RAID array with a ZFS– and Linux-based product is more flexible in terms of configuration, and it sets you free from 3rd-party hardware products or additional tools to manage them. Furthermore, Open-E JovianDSS allows you to run helpful diagnostics to prevent unexpected failures.
That’s a lot to consider, especially if your budget is limited, and we know that. But don’t worry – if you wish to count the expenses focusing on the abovementioned factors, don’t hesitate to use our Open-E JovianDSS Storage and RAID Calculator. That makes it vastly easy to choose relevant solutions tailored to your business.
Leave a Reply