Flash storage is reshaping enterprise IT with QLC and NVMe innovations, but HDD still dominates on cost. Explore flash vs HDD, cloud storage, and the future of all-flash datacentres.
The past year has been pivotal for enterprise storage. Flash storage – once considered too expensive for many workloads – has pushed into new areas, thanks largely to the rise of quad-level cell (QLC) technology, which delivers higher density and a lower cost per gigabyte (GB). For the first time, flash has started to compete seriously with nearline hard disk drives (HDDs).
At the same time, the flash market has seen sharp pricing fluctuations, with costs dropping towards HDD levels before rebounding as manufacturers adjusted supply. Some industry players even predict the hard drive’s imminent demise and the eventual dominance of the all-flash datacentre.
So, where does enterprise flash storage stand today? In this article, we explore what flash is, the role of QLC and NVMe, how it compares to HDD in cost and performance, the use of flash in cloud environments, and whether spinning disks still have a future.
What is enterprise flash storage?
Enterprise flash storage refers to arrays made up of multiple flash drives, usually housed in rack-mounted enclosures in the datacentre. These arrays aggregate the capacity of dozens – or even thousands – of drives, controlled by dedicated hardware that manages input/output (I/O) from hosts, allocates data, and performs maintenance tasks such as:
- Wear levelling – ensuring data writes are evenly spread to prolong drive life.
- Garbage collection – reclaiming space by consolidating fragmented data.
- Error correction – protecting against bit errors inherent in flash media.
Enterprise flash arrays can scale from tens of terabytes (TB) to multiple petabytes (PB). Just like HDD-based systems, they provide block, file, or object storage, depending on workload needs.
What is QLC flash storage and why does it matter?
QLC stands for quad-level cell flash, where each memory cell stores four bits of data using 16 distinct states. By comparison, triple-level cell (TLC) stores three bits, while older technologies like multi-level cell (MLC) and single-level cell (SLC) stored two and one bit, respectively.
The benefit of QLC is clear: higher density at lower cost. This makes it attractive for workloads where capacity matters more than endurance – such as unstructured data storage, analytics, or archiving.
The trade-off, however, is endurance. More states per cell increase the risk of errors and reduce the number of times data can be reliably rewritten. As of early 2024, most enterprise arrays still rely on TLC for mission-critical tasks, but QLC is gaining ground rapidly for secondary and nearline workloads.
What is NVMe and how does it improve flash storage?
Before NVMe, flash drives often used protocols designed for spinning disks, like SATA and SAS. These worked, but they didn’t take advantage of flash’s low-latency potential.
Non-volatile memory express (NVMe) was designed specifically for solid-state storage. Its key innovations include:
- Optimised I/O queues – supporting thousands of parallel operations versus the limited queues of SATA/SAS.
- Reduced latency – cutting access times from milliseconds to microseconds.
- Scalability – NVMe-over-Fabrics (NVMe-oF) extends flash performance across networks using Ethernet, Infiniband, TCP, or RDMA.
NVMe is now the default choice for performance-hungry workloads, from databases to AI training environments.
What role does HDD still play?
Hard disk drives remain the workhorse of many datacentres. Based on spinning platters and magnetic read/write heads, HDDs are slower but far cheaper than flash. They are widely used in hyperscaler datacentres, where cost-per-gigabyte dominates decision-making.
Enterprise HDD arrays still provide petabytes of storage at relatively low prices, particularly for cold or nearline data. High-capacity HDDs are available in sizes up to 22TB, with roadmaps promising even larger drives.
How does flash performance compare to HDD?
The performance gap is dramatic:
- Latency: Flash offers microsecond-level access times, versus several milliseconds for HDD.
- IOPS: Flash arrays deliver many times the IOPS of HDD, making them ideal for transactional workloads.
- Throughput: Flash drives achieve gigabits per second (Gbps) of throughput, four to five times faster than HDD.
The lack of moving parts gives flash a huge advantage. HDDs are limited by physical mechanics, while flash is purely electronic.
In short, flash is orders of magnitude faster – but still more expensive per gigabyte.
How do flash and HDD compare on cost?
Cost remains the key battleground. In late 2023, flash prices dropped close to HDD levels, only to spike again as suppliers reduced output. By September 2024:
- Flash averaged $0.085/GB, down from a high of $0.095 in April 2024.
- HDD averaged $0.039/GB for SAS/SATA drives, with SATA as low as $0.035/GB.
That means flash still costs roughly double HDD on a per-gigabyte basis, although the gap has narrowed over time.
For a 20TB deployment:
- Flash would cost ~$1,700 in September 2024.
- HDD would cost ~$780 for the same capacity.
While flash is becoming more affordable, HDD continues to dominate where massive, low-cost capacity is needed.
Will flash replace HDD entirely?
Predictions about the death of HDD have circulated for years. Some vendors, like Pure Storage, boldly claim spinning disks will be obsolete by 2028, pointing to advances such as 300TB flash modules expected by 2026.
Others are more cautious. Vendors like Panasas note that hyperscalers still rely heavily on HDD – sometimes with 90/10 HDD-to-flash ratios – because the economics still strongly favour disk for bulk storage.
The reality is likely to be hybrid. Flash will dominate performance-sensitive workloads, while HDD will remain viable for high-capacity, low-cost archives.
How does flash work in cloud environments?
Cloud providers allow customers to specify performance tiers, often without needing to know whether the underlying media is flash or HDD. That said, all major hyperscalers – AWS, Microsoft Azure, and Google Cloud Platform – offer explicit flash-backed options, with tiers based on IOPS and throughput guarantees.
Examples include:
- AWS EBS: SSD-backed general-purpose and provisioned IOPS volumes.
- Azure NetApp Files: Enterprise-grade flash file storage.
- Google Persistent Disk: Balanced and extreme performance SSD options.
Cloud flash storage is often priced by performance tier, giving customers flexibility to match cost and performance requirements.
What’s next for enterprise flash?
Looking ahead, flash adoption will continue to expand into new areas:
- QLC drives will become the default for unstructured workloads.
- TLC and higher-end NVMe will remain critical for transactional systems.
- NVMe-over-Fabrics will extend flash performance across entire datacentres.
- Flash in the cloud will continue to blur the line between on-premises and hosted storage.
The “all-flash datacentre” may not happen as quickly as advocates predict, but the trajectory is clear: flash is increasingly dominant in enterprise storage strategies.
The takeaway
Flash storage is no longer limited to high-performance niches. With QLC bringing down costs and NVMe delivering blistering speeds, flash has moved firmly into mainstream enterprise storage.
However, cost remains a sticking point, and HDD still plays an essential role in large-scale, low-cost capacity. While the all-flash datacentre may eventually arrive, in the near term, most organisations will continue to run hybrid environments where flash and HDD coexist, each serving its optimal use case.
Read more about storage media
File, block and object storage: How do they work in the cloud era? File, block, and object storage remain core to data management in the cloud era. Learn how they differ, how they evolved, and where each model fits best.
S3 Storage: What it is, how it works, use cases. Discover how AWS S3 object storage works, its classes, use cases, and S3-compatible on-prem options that bring cloud-scale storage to hybrid environments.