The flexibility of choosing storage options in today’s business landscape and the stability of these media has created a buzz around many enthusiasts in the storage industry. The SSDs have become a part of the enterprise storage since their advent in the industry. These drives differ from mechanical drives in many respects, such as cost, capacity, performance, and reliability. All these factors come into play when these drives are chosen for enterprise applications. However, the most important factor from the perspective of enterprise storage is reliability.
Identifying the Safety of Data on Storage Media
Mechanical drives are long known to slow down your system’s processing speed. With the introduction of solid-state drives, it was realized that combining the NAND flash with the existing HDD technology can increase the performance of a system many folds. However, the objective of achieving an optimized computing experience cannot satisfy the reliability needs of enterprises. When it comes to securing your data, the reliability of storage media is the primary requisite.
The technology used in SSDs is relatively new with respect to that of mechanical drives. SSDs now use flash memory that is manufactured at 25 nm. Earlier SSDs were manufactured with 3x nm MLC NAND that allowed for nearly 5000 P/E cycles without any effect on the data stored. The 25 nm NAND has resulted in lower number of P/E cycles, subsequently increasing the risks of early write
exhaustion. There are a number of other factors that can affect the ability of your drive to retain data, such as compression, garbage collection, and write amplification. When comparing their reliability to HDDs, the lack of moving parts may be a plus for SSDs. However, this does not bring us to the conclusion that SSDs have some sort of reliability edge over HDDs.
Why is Reliability Critical
The SSD manufacturers are trying to make two promises to the consumer, i.e. high reliability and high performance. The feedback from several big data centers sheds light on the fact that SSDs do fail, however, the failure rates are lower than those of HDDs. This implies that if SSD’s reliability is not a certain advantage over the HDDs, then the only reason to go for these drives is performance. The
performance of both low-end and high-end SSDs is greater than that of HDDs.
SSD performance has been following a steep growth path and will continue to do so in the foreseeable future. This manifests in itself that other factors may be considered when differentiating these drives from mechanical drives. This indeed turns the focus to reliability. These SSD drives will remain a maturing segment of the industry as long as they keep confronting issues.
Examining Reliability and Failure Rates
There are several factors that may affect the reliability of hard drives and SSDs in several ways.
This can be attributed to their different architectures. A hard drive only comprises moving parts, which despite of their tolerances are doomed to fail at some point. On the contrary, SSDs employ a completely different architecture that eliminates any type of failures due to wear and tear of mechanical components. With SSD storage, you won’t expect the read/write head to crash untimely or the motor
to wear out over time. However, there are other factors that determine the reliability quotient of SSDs.
Since you do not know where the static LBA space resides on an SSD, they are prone to failures. Firmware is one more reason that can lead to problems with SSD. Moreover, different methods used by vendors and manufacturers to report reliability can create a serious difference in the perception of reliability. The real truth behind the reliability story may be revealed over the years to come.
The Need to Simplify Storage Configuration
The failure rates of drives differ with drive models, vendors, and manufacturers. However, this difference is very small when seen from a rational perspective. The reliability and performance of storage media are the two major factors on which all enterprise applications overly depend. In order to meet their throughput needs, the enterprises use multitude of short-stroked high RPM disks in RAID. They end up deploying more mechanical storage in servers in order to scale up. If these drives are replaced by a handful of SSDs, you may wind up having a simplified storage configuration with reduced power and cooling requirements.
Reducing the number of mechanical drives in your array would imply better reliability through fewer devices. Having a 1:4 or a similar swap would let you increase your performance substantially by spending a lot less.
Concerns for Desktop Users
From the perspective of enthusiasts, it doesn’t make sense to claim that an SSD is more reliable than a mechanical drive. Given the number of firmware bugs and write endurance issues with SSDs, it is yet to be evolved into a robust and reliable technology. A piece of hardware may fail anytime, regardless of whether it has moving parts or not. It is not feasible to deliver products that completely alleviate the risk of failure. This fundamental rule seems to hold true for both SSDs and HDDs.
Reliability is critical to enterprise storage. Several researches and studies came clean on the fact that SSDs do have the benefit of additional reliability over hard drives. However, there is no definitive conclusion to this as SSDs are in the market for no more than two years. These researches and studies are based on the data and statistics collected over these two years. There is nothing with us beyond these years that can support this very claim.
Hard drives have been around for hundreds of years and the reliability tests for these drives utilize the data that is well documented and verified. This makes us pretty confident that these mechanical drives will continue to show the same failure rates as they did in the past. With that said, there is no arguing to the fact that the question still remains unanswered. On the flip side, one may fight these drive failures by choosing to keep all mission-critical data in multiple places.