- Hard drives are getting bigger but performance is a problem in data centers
- Meta looks QLC Flash as a solution that bridges the gap between HDDs and TLC
- With 512TB QLC SSDS planned, Meta switches away from E1.S and E3
Hard disk drives are still choosing go-to storage for most data centers, mainly because they are cheaper and use less power than TLC flash.
But while HDDs are getting bigger, their input/output performance has not kept going. As a result, bandwidth per Terabyte shrunk, forcing data centers to move often available data to TLC Flash or Overprovision Storage.
In a recent Facebook engineering blog post, Meta outlined its approach to integrating QLC Flash as a solution to these challenges. QLC Flash has been around since 2009, but the adoption has been slow due to limited capacity, high costs and lower writing endurance.
QLC SSDs are the future
Meta explained that it sees QLC as the middle ground between HDDs and TLC SSDs. The technology provides sufficient benefit to workloads that have previously been dependent on 16TB and 20TB HDDs and supports large Batch I/O tasks that need more than HDDs can offer. Since most power consumption in flash storage comes from writing, QLC’s lower writing activity makes it a possible, energy efficient option.
While QLC is currently cheaper than TLC, it is not yet a price point for widespread implementation. The power savings make it an attractive opportunity for certain workloads, and the recent progress makes the flash format more practical. The arrival of the 2TB QLC NAND door and the widespread use of 32-die stacks drives higher storage density, and QLC SSDs are expected to scale faster than TLC. This will eventually help improve server efficiency while lowering costs.
Meta plans for QLC SSDs to reach capacities as large as 512TB and see E1.S- and E3 form factors that died on arrival at its needs. Instead, it intends to focus on U.2-15mm and Pure Storage’s Directflash module to scaling QLC in its infrastructure.
Meta also refines its storage software to handle QLC’s high density. Since QLCSSDs have a large space between reading and writing speeds, proper planning is needed to maintain performance.
As the QLC technology progresses, Meta expects its role in data centers to grow, a point of view shared by Storageerview.
The site’s Brian Beeler writes: “Our test has strengthened QLC’s role in AI storage. A recently deeply dive explored how Solidigm’s D5-P5336 QLC SSD performs in control points workloads for AI model training. These tests showed that although TLC SSDs are still leading in writing environments, QLC SSDS keeps their own in terms of capacity, efficiency and reading of performance – making them a strong opportunity for AI pipelines. “
