MENU

Hybrid Versus All-Flash Storage Arrays for Enterprise Storage

When flash storage first emerged onto the scene, its use by enterprises was limited due to its cost. However, increased demand for flash storage over the years caused manufacturers to increase production. It turned out that vendors overestimated the future market value of flash storage, resulting in oversupply and lowered prices. This made all-flash arrays a popular choice for certain enterprise storage applications—but not for all. Even with the multiple benefits offered by all-flash storage compared to older technologies, most IT systems don’t use a single or homologous type of storage solution. While many enterprises choose hybrid storage arrays over all-flash because of price, hybrid arrays offer much more than monetary savings.

Hard Disk Drives (HDDs) Versus Solid-State Drives (SSDs)

In order to fully comprehend how a hybrid storage array compares to an all-flash storage array, it’s necessary to go over the differences between HDDs and SSDs.

HDD

Traditional hard disk drives use magnetic heads that read and write digital data to rapidly rotating platters (or disks) coated with magnetic material. HDDs are classified as a type of non-volatile storage, meaning that they retain stored data whether or not they have power. HDD technology was first introduced by IBM in 1965 and has since then been the principal secondary storage medium for computers as far back as the 1960s. Even with the growing popularity and falling prices of SSDs, HDDs are still used in many computer devices even today due to the vast amount of data they can store per gigabyte.

One of the downsides of HDDs is their relatively lower performance when compared to SSDs—only offering approximately 75 to 200 input/output operations per second (IOPS) and an average rotational latency of between seven milliseconds (4,200 RPMs) and two milliseconds (15,000 RPMs). Traditionally, IT departments increased storage array performance by adding more HDDs (often referred to as increasing the number of spindles by IT professionals) and short stroking the disks which involved using only the outer portion of the rotating platters.

Of course, short stroking a disk means less of the potential disk capacity is used—the benefits of short stroking can only be seen with very small drive sizes, offering just around 40 percent better random seek time and slightly better bulk transfer rates. When one considers that short stroking requires specialized software and consumes a lot of power when used in RAID 0 arrays, using SSDs might actually seem a better solution.

SSD

SSDs contain flash memory through a series of integrated circuit assemblies rather than a rotating disk. Since SSDs don’t use disks, they’re much more shock resistant, use less power to operate, and offer faster access times with lower latency. Instead of magnetic tracks, SSDs store data in one to four bit-capacity (one to four bits per cell) semiconductors.

In terms of industrial applications, single-level cell (SLC) SSDs use only one cell and are more reliable, more durable, and less prone to errors. SLC SSDs are normally pricier than other SSD devices. Multi-level cell (MLC) SSDs have two cells and are also used in a number of industrial applications. MLC SSDs are slower than SLC SSDs because it requires slightly more time to write two bits into a cell rather than one. Since MLC SSDs write data to the NAND flash more frequently than SLC SSDs, they’re less durable and reliable.

Triple-layer cell (TLC) SSDs are comprised of three cells. For the most part, TLC SSDs remain the most commonly used type of SSD for several reasons. Though they offer less speed, reliability, and durability than SLC and MLC SSDs, they have more capacity since they write three bits per cell. In terms of price, TLC SSDs cost relatively less than the first two types. Lastly, TLC SSDs may be less durable than the latter two SSD types but they normally last several years or so.

Quad-level cell (QLC) drives write four bits per cell, which has a significant effect on their overall performance. It used to be that QLC SSDs ran out of cache capacity during sizeable file transfers (usually 40 GB or more). However, manufacturers are working on ways to optimize QLC SSDs so this is no longer an issue. Even then, budget-level QLC SSDs only offer 100 terabytes written (TBW) on 500 GB devices and 200 TBW on 1 TB—not good for durability.

As one can observe, choosing an SSD type for an all-flash storage array isn’t as clear and cut as it might first seem.

Solid State Hybrid Drives (SSHDs): Merging Speed With Capacity

SSHDs (also referred to as hybrid drives) are storage appliances that combine the fast speed of SSDs with the capacity of HDDs. In SSHD systems, the SSD retains the most frequently used data which improves overall performance by acting as a cache for the HDD. SSHD products usually integrate SSD NAND flash memory and HDDs into a single device. Fundamentally, SSHDs identify data components most directly linked to performance and store them in the NAND flash memory.

It’s important to understand that the term solid-state hybrid drive is more accurate than the generic term hybrid drive, which used to define both SSHD devices and non-integrated combinations of SSDs and HDDs. An example of the latter is a laptop with dual-drive systems—separate SSD and HDD components within the same 2.5-inch HDD-size unit. Both components are visible and accessible to the operating system as two separate partitions, whereas an SSHD is a smaller SSD and larger HDD soldered together to make a single device.

Hybrid Versus All-Flash

When it comes to deciding a hybrid storage array or an all-flash storage array, it will largely depend on the type of transaction-processing system (TPS) is at play. Hybrid storage arrays were developed to increase IOPS while decreasing latency. By merging SSD NAND flash with HDD technology, you have one foot in the best of both worlds.

While this is a definite improvement over HDD arrays, performance statistics aren’t always clear or conflicting, causing IT managers to second guess investing in hybrid storage solutions. For most, the lower price per gigabyte has always made hybrid storage arrays the most popular choice.  Besides being more economical, here are several other things that make hybrid a better option for some enterprises.

  • Much better performance than an all-disk storage solution
  • Provides the best of two technologies—speed and capacity
  • A great choice for situations where capacity is the top priority
  • A great choice for file storage, backup, and recovery
  • A very suitable solution for low to moderate transaction environments
  • Many hybrid solutions offer hundreds of thousands of IOPS

Just as improvements on the SSD design continue, manufacturers are improving HDD technology as well. Groundbreaking ultra-high capacity HDDs with microwave-assisted magnetic recording (MAMR) technology should facilitate hard drives with 40TB of capacity by 2025. With that said, hybrid storage arrays will remain a competitive option to all-flash storage arrays.

High-performance computing, big data, and cloud applications require affordable solutions that provide great performance. With the ARI-400 Series, you get declustered RAID greatly reducing rebuild times and a dual controller system with 7 Gbps reads and 5.5 Gbps writes—sustained. In addition to that, the ARI-400 Series with hybrid arrays can achieve up to 600,000 IOPs per second output. To learn more about the ARI-400 and other hybrid storage array solutions offered by RAID.Inc, contact one of our friendly and knowledgeable experts today.