Micron Technology Inc. said it is now shipping its 245 terabyte capacity 6600 ION solid state drive (SSD) targeted at improving the rack-scale storage density for data centers, specifically, AI, cloud, enterprise and hyperscale workloads.
According to Micron, the SSD requires 82% fewer racks to achieve the same raw storage capacity compared to hard disk drive (HDD)-based deployments.
The SSDs are built using Micron’s G9 QLC NAND technology. This allows users to store and process significantly more data in far less space while simultaneously reducing power and cooling demands for large-scale, data-intensive workloads.
“AI workloads are driving massive growth in shared data, continuing the shift of data center storage share from HDDs toward SSDs,” said Jeremy Werner, senior VP and GM of Micron's Core Data Center Business Unit. “With 245 TB in a single SSD, the Micron 6600 ION makes solid state storage the clear choice for modern data centers. This breakthrough capacity gives data center operators a critical new lever to improve rack-level total cost of ownership, especially as power availability becomes a defining constraint for AI infrastructure scale.”
Why it matters
AI data centers are pushing capacity limits. If it is possible to fit more storage into every rack inside a data center it allows for less power, less floor space and less operational overhead, Micron said.
With larger SSDs, it is a reduction in total cost ownership for data centers while providing more efficiency for extreme-capacity deployments in AI workload performance and energy savings against traditional storage options, the company said.
Micron said in its testing, the 6600ION SSD demonstrated gains in energy, throughput and latency against HDDs. Specifically for AI workloads, this means:
- Up to 84 times better energy efficiency
- 8.6 times faster AI preprocessing
- 3.4 times better ingest throughput
- Up to 29 times lower latency
In object storage workloads, this means:
- Up to 435 times better throughput per watt
- 96 times faster time to first byte
- 58 times better aggregate throughput
