Is hybrid storage heading toward extinction? This question is being asked as the direct result of the rapid growth of all-flash arrays as well as some highly hyped initial public offerings by AFA vendors, leaving IT storage administrators to try to make sense of it all. The question may seem simple; however, the answer is anything but.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
It starts with a definition: What exactly is hybrid storage? Conventional wisdom currently defines a hybrid storage array -- also known as a hybrid flash array -- as a storage data management system that manipulates memory using a combination of flash, such as solid-state drives (SSDs), and slower hard disk drives (HDDs) to provide the most cost-effective, high-performance I/O. Hybrid storage attempts to balance the cost of the storage medium with the I/O performance required. Flash can be set up as a read cache, read and write cache or as a high-performance storage tier that captures all writes and moves the data to HDDs based on policies such as data age, access frequency or data importance.
The rapid increase in flash SSD capacities and an equally rapid decline in pricing per terabyte is placing market pressure on HDDs as a viable storage medium going forward. The price gap between high-capacity flash SSDs and high-capacity HDDs is narrowing much faster than was previously anticipated due to significant gains in flash NAND 3D technology.
The highest capacity 2.5-inch small form factor (SFF) flash SSD that was realistically available in calendar year 2015 was approximately 4 TB. In early 2016, SFF flash SSD capacities were approximately 16 TB and will reach approximately 32 TB by the end of the year. Compare that to 3.5-inch large form factor HDDs that topped out at 10 TB in 2015 and 2016.
Problems at the heart of hybrid storage arrays
There's a catch with high-capacity flash SFF SSDs. Many higher capacity flash SSDs will be based on triple-level cell (TLC) NAND technology. TLC is definitively the lowest cost flash technology today, but it has some significant issues, such as a relatively short wear life. Remember that flash SSD reads have no cost to the life of the drive, whereas writes do. Each write destroys some of the storage material. Wear life is the number of program/erase (PE) cycles; TLC has at least an order of magnitude lower number of PE cycles than multi-level cell (MLC) flash.
TLC writes and reads are also measurably slower than MLC, though they still have at least 10 times the performance of HDDs. What that means is that TLC 3D NAND-based SFF SSDs are more suitable for reads than for the high-intensity writes typically associated with flash SSDs. The most suitable workloads for this type of low-cost, high-capacity, high-density, slower flash SSD is the same as that for the HDD currently used in a hybrid storage array.
The hybrid storage array is evolving from a combination of memory, flash SSDs and high-capacity HDDs to a blend of memory; high-performance, write-optimized flash SSDs; and high-capacity, read-optimized flash SSDs. Another way of thinking about the future of hybrid storage is as a multi-tiered all-flash array. The principles and logic behind hybrid storage are the same -- not all data has the same value, and the value of that data should be aligned with the cost of the storage media upon which it is stored.
So while hybrid storage with HDDs is a dying breed, we can look forward to the long life of hybrid multi-flash storage.
Hybrid arrays pushed aside by all-flash
Virtual Volumes from vSphere helps hybrid storage and AFAs
Everything you need to know about hybrid flash arrays