![Industry Perspectives Industry Perspectives](https://eu-images.contentstack.com/v3/assets/blt8eb3cdfc1fce5194/blt79a1824cfad941c4/664ba129be90a086233eb8e0/Industry_Perspectives_DCK_light.png?width=700&auto=webp&quality=80&disable=upscale)
Insight and analysis on the data center space from industry thought leaders.
All-Flash vs. Hybrid Flash – Choosing What's Right for You
When it was just emerging, flash storage was expensive and used in limited fashion, but prices have dropped significantly, lowering that barrier to entry.
March 5, 2019
Nitin Reddy Tandra is Senior Storage Engineer at Broadcom.
With recent developments in file storage arrays, flash prices dropping significantly and many competitive vendors offering different types of all-flash storage arrays to meet Network File System/SMB performance demand in IT, the flash storage decision-making process has grown quite complex for IT managers and professionals.
The Rise of All-Flash
Flash storage has seen rapid growth with its ability to improve performance and provide the same capacity as hybrid flash in a smaller footprint. Flash storage allows IT organizations to leverage benefits like reduced power consumption, physical rack space consumption, and power and cooling reduction over hybrid flash solutions.
During the initial days of its emergence, flash storage was expensive and used in limited fashion, such as for a subset of performance-driven applications only. But flash prices have dropped significantly, lowering that barrier to entry.
Let’s examine both use cases.
Hybrid flash arrays: Hybrid flash arrays are a combination of SSDs and HDDs. Hybrid arrays allow organizations to take advantage of flash’s high performance and optimization levels, while leveraging the benefits of HDD capacity.
Why hybrid flash: Hybrid arrays were designed to increase Input/Output Operations Per Second (IOPS) and decrease latency by mixing flash with disk. There is no doubt that there is performance improvement over using pure HDDs. But because performance numbers vary for hybrid solutions using different types of workloads, IT managers and professionals shouldn’t expect to see exact numbers across the board. Organizations that have storage infrastructures dependent on HDDs find that adding some flash storage either for storing metadata or serving data for certain business critical applications makes more sense and can drive application performance needs. It also allows organizations to slowly transition from existing legacy equipment, which achieves improved performance levels while not biting off the entire cost of going all-flash all at once.
Hybrid flash advantages: Hybrid solutions offer increased adaptability and efficiency. Traditional HDDs can typically store the largest amounts of data. But that has a downside: It makes storage array performance drop drastically. By mixing the speed of flash with the capacity of HDDs, the hybrid approach offers a balanced infrastructure between performance and capacity.
Hybrid flash is more affordable when users compare cost per gigabyte and highly suitable for low- to mid-range transaction environments. It can also be seen as the best of both worlds: higher performance than an all-disk solution and cheaper than all-flash.
Hybrid flash disadvantages: Hybrid flash performance can’t match an all-flash solution. However, depending on company needs, hybrid storage may suffice as many modern solutions offer up to thousands of IOPS. But in some cases, applications compete for the flash tier of storage, which can lead to overcrowded flash and reduced performance for applications. And since users end up mixing SSD and HDD, there is a need to manage automatic-tiering – which is not 100 percent effective: Automatic-tiering design cannot keep up with application workloads that are constantly changing. In addition, hybrid flash still takes up lot of rack space, power and cooling, so users end up paying more anyway.
Why all flash: All flash arrays are a perfect fit if a main concern is high performance within a smaller footprint solution. With data stored on solid state disks or NAND chips, this type of solution can provide millions of IOPS.
All-flash advantages: All-flash systems are composed of deduplication and compression to address high capacity requirements. For example, users can fit a petabyte of usable data in an 8U rack solution generating ~3-4kWh power per month using all flash. By comparison, a typical high-end hybrid flash system can take close to 100U of rack space to accommodate a petabyte of usable data without dedupe and compression and generates ~19-20 kWh of power per month.
All flash is based on SSDs/NAND with no moving parts: the systems use flash memory to quickly read/write and perform I/O operations. The performance benefit alone makes it worthy to run mission-critical apps on all-flash — where faster response time for read/write/delete can make a huge bottom-line difference. All-flash also leverages fast 40G Ethernet switches, which has become a norm.
But performance isn’t the only advantage. There is no tiering involved since it’s all flash, so there is much less operational work. In addition, file storage backups complete in a faster time, which enables the user to meet any service level agreement (SLA) with an internal customer.
Finally, all-flash arrays are typically smaller and take up less space on the storage rack. Also, because flash runs on SSDs, it uses less power as there are no moving parts — and less cooling due to fewer components producing heat. In fact, there's a 75-80 percent reduction (for 100 PB of data) in rack/power and cooling using an all flash solution vs. a hybrid solution, which in today’s world is huge given the costs for racks/power/cooling incurred by an organization to maintain it.
All-flash disadvantages: All-flash is expensive when you’re looking at raw numbers, but dollars/IOPS is where all flash solutions really have the upper hand – and it is what tends to persuade IT managers and professionals to go for an all flash solution.
The bottom line: Only the individual and the IT group truly knows the organization’s present storage environment and can project future needs based on objectives and financial budgets. There’s no one-size-fits-all solution. To make an informed and strategic storage decision, be sure to characterize current storage challenges, make sense of explicit remaining tasks at hand (current and future) – and characterize essential concerns (execution, limit, cost).
Opinions expressed in the article above do not necessarily reflect the opinions of Data Center Knowledge and Informa.
Industry Perspectives is a content channel at Data Center Knowledge highlighting thought leadership in the data center arena. See our guidelines and submission process for information on participating.
About the Author
You May Also Like