- Hard discs are getting bigger, but performance is a problem in data centers
- Goal VE QLC Flash as a solution, closing the gap between HDD and TLC
- With 512TB QLC SSDS planned, goal is moving away from E1.s and E3
Hard disk units remain the storage option for most data centers, mainly because they are cheaper and use less power than TLC Flash.
But while HDDs are becoming larger and larger, their input/output performance has not been maintained. As a result, the bandwidth by Terabyte has been reduced, which forced data centers to move frequent access data to the FLASH TLC or overcrowded storage.
In a recent Facebook engineering blog post, Meta described his approach to integrate QLC Flash as a solution to these challenges. QLC Flash has existed since 2009, but adoption has been slow due to limited capacity, high cost and lower writing resistance.
QLC SSDS are the future
Meta explained that he sees QLC as the average soil between HDDs and TLC SSDs. Technology offers sufficient performance for workloads that were previously based on HDD of 16 TB and 20TB and admits large lots of lots that need more than HDDs can offer. Since the majority of energy consumption in Flash storage comes from writings, the lowest QLC writing activity makes it a feasible and efficient option in energy.
While QLC is currently cheaper than TLC, it still has no price for generalized implementation. However, energy savings make it an attractive option for certain workloads, and recent advances are making the flash format more practical. It is expected that the arrival of the 2TB QLC NAND die and the widespread use of 32 batteries of 32 die are promoting a greater storage density and the QLC SSDs are climb faster than TLC. This will finally help improve server efficiency while reducing costs.
Meta is planning that QLC SSDs reach as large as 512TB and see the factors in E1. and e3 as dead when the arrival of their needs. On the other hand, it intends to focus on U.2-15 mm and the Pure Storage directflash module to climb QLC in its infrastructure.
Goal is also refining its storage software to handle the high density nature of QLC. Since QLC SSDs have a large gap between reading and writing speeds, adequate programming is needed to maintain performance.
As QLC technology progresses, goal expects its role in data centers to grow, a vision shared by Storagereview.
Brian Beeler of the site writes: “Our tests have reinforced the role of QLC in the storage of AI. A recent deep immersion explored how the QLC QLC D5-P5336 of solidigm works in the work points workloads for the training of the AI model. These tests showed that, although TLC SSDs still lead in writing environments, QLC SSDs remain their own with respect to the capacity, efficiency and reading performance, which makes them a solid option for AI pipes. “
