HDDs are one of the oldest forms of computer storage and have been in use for over 60 years. They were first developed by IBM in 1956 (they were bigger than a fridge) and have since been improved on dozens of times to the point of where we’re at today.
In today’s market, HDDs are touted for their low price per GB and extended lifespan when compared to a solid-state drive. Today, SSDs are not only affordable but commonly found in consumer electronics as well. SSDs still cost more than HDDs if you break it down into price per GB, but there are clear advantages to owning an SSD that often make them worth the extra cost
While it might be tempting to replace all the hard disk drives in a system with solid state drives (SSDs), the flaws with large capacity SSDs become apparent.
California-based Nimbus Data intends to launch what will be the largest-capacity solid-state drive: a 100-terabyte SSD called the ExaDrive DC100. In addition, the drive will be accompanied by a smaller sister drive at 50-terabytes.
The drives would provide enough flash memory for 20,000 HD movies or 20 million songs. While the new drive can support up to 100,000 reads/write operations per day and up to 500 MB per second of throughput, the device will be available to customers with a guarantee of five years. Nimbus Data is coining the drive under the phrase, "Unlimited Endurance."
However, will the drive be as "unlimited" as Nimbus Data says it will be?
Being the forefront of capacity in the world of SSDs doesn't exclude you from the issues that follow with a drive of such magnitude. Despite a large amount of storage being available in a small form factor, is it practical to utilize an SSD in big data architecture?
While it might make sense to utilize SSDs in areas where server space is unreasonably expensive; for most use cases it will be better to utilize a number of smaller drives instead, that will decrease density, but will increase performance, reliability and data availability.
For instance, 100-terabytes is a lot of data to lose, and a lot of data to restore in case of a failure until a replacement drive is populated and ready for service. Additionally, the possibility for data loss increases with the increasing capacity. Huge capacity raises the number of components, therefore decreasing reliability. P/E cycle of the drive could also be greatly affected due to the huge capacity.
A FIFO pattern might do well to extend media life as close to the theoretical maximum. But imagine a situation where a substantial amount of the data is permanently stored and those blocks of memory are essentially out of circulation, the drive will now have less memory to service the promised 43 TB of writes per day, increasing wear on the blocks that are in circulation.
That can actually be mitigated by moving stale data around, which will also be beneficial to avoid running into issues with data retention, although this will increase complexity and possibly decrease performance. But without that scheme implemented, wearing the drive out using the promised "unlimited endurance" is actually quite possible.
The cost of the 100-terabyte SSD would far exceed any alternatives that are available now.
The next biggest SSD, the Seagate 60-terabyte SSD, costs approximately $7,000 dollars, meaning the ExaDrive DC100 could reasonably cost around $10,000 dollars.
However, an alternative to purchasing one 100-terabyte SSD would be purchasing 11 10-terabyte Seagate Ironwolf HDDs for around $5,000. The price to buy HDD is significantly cheaper and comes with the peace of mind that if one drive fails it can be replaced with an affordable alternative. An even cheaper option would include 35 3-terabyte WD red HDD for less than $4,000 dollars.
But all this changes with Intel's new design.
Intel might be providing a better option than simply using tons of hard drives or one large SSD. Intel has introduced its new form-factor for server-class SSDs. The new "ruler" design is based on the in-development Enterprise & Datacenter Storage Form Factor (EDSFF) and is intended to enable server makers to install up to 1 PB (1 petabyte = 1000 terabytes) of storage into a single storage server.
The ruler form-factor was developed specifically for server drives and is tailored for requirements of data centers. As Intel puts it, the ruler form-factor “delivers the most storage capacity for a server, with the lowest required cooling and power needs”.
Additionally, Intel's ruler SSD will be offered with Intel’s enterprise-grade 3D NAND memory and a proprietary controller. Meaning that the security and reliability of Intel SSDs will be accompanying a forefront to the SSD market.
The Intel ruler could potentially be incorporated into a system that far exceeds the capacity of any previous systems.
For instance, with Trenton Systems' shift towards data management, it is possible that the same techniques to achieve multiple GPUs in one system could be applied to multiple Intel Ruler SSDs.
Having multiple petabytes of storage (that is, segmented storage that is scalable) would re-imagine the way storage is handled, and the risk is minimized since each 'ruler' can hold a fraction of your data vs one large SSD holding everything you need.
In the end, the prospects of SSDs are being realized with devices such as the Nimbus data 100TB SSD and Intel's Ruler SSD.
The future of large capacity SSDs are finally being implemented and could potentially trickle its way down to consumers, thus promoting a new market of larger and newer SSDs that will outperform previous models.