A solid storage hardware solution is often elusive. We’ve discussed why hard drives fail so often and how X-IO has addressed hard drive failures with their ISE series.
In addition to mitigating the primary sources of failures for hard drives, the ISE chassis also treats drives in a manner to optimize their performance.
The X-IO ISE can’t bypass physics, not directly anyway, but it does treat drives in a manner that optimizes their physical characteristics and gives the drive its best performance. For example, the 20 percent spare space reserved within an ISE enclosure comes from the innermost tracks of the drives, which means that it is always using the fastest outer tracks of the drive and will only use the slower inner tracks if necessary for recovery.
Additionally, within those outer tracks, the ISE lays out data in a proprietary manner called RAGS which writes to that 80 percent of the drives in a uniform manner. This means that performance is consistent and stable from the first byte to 100 percent capacity, something that no one else in the industry achieves. This RAGS layout also allows for very granular and intelligent recovery within the Datapac which greatly increases recovery times, and minimizes the impact of recovery to a negligible level.
Find out how physics can impact a hard drive’s performance.
The ISE chassis has a very homogenous and controlled environment, which is what facilitates all of this advanced behavior. Within each drive enclosure, the ISE always knows exactly how many drives there will be, and what the RAID layout will need to look like. Because of this, a large portion of the RAID protocol can be pruned out, leaving only what is necessary to facilitate the specific environments available.
The ISE also does not have to be able to accommodate many vendors and many sizes of drives, all of the code is specifically written for the drives that are sold in Datapacs. Knowing that the drives will always exist in an ISE also allows the firmware on the drive to be stripped down to exactly what is necessary to function in that environment, which means there isn’t superfluous code.
Read up on how generalization can slow down hard drives.
The self-contained nature of an ISE system allows for it to scale infinitely and linearly in performance. The controllers within an ISE are never expected to service more than 80 drives. This means that as an extra drive enclosure is added, additional cache, additional interconnects and additional processing power is also added, which means that the performance will keep scaling indefinitely.
In other words, there will never be a point where suddenly, to add another 20 drives, you need to forklift the entire system you currently have to add additional capacity or performance.
Find out how improper interconnectedness between drives can slow them down.