May 16th, 2018 by Adam Armstrong
Dell EMC Takes A Stab At 1PB/1U With High Density M.2 Sleds
Dell EMC has announced a potential fix to the issue of getting 1PB of density into a 1U space. Dell EMC is looking to leverage low cost M.2, SSD solutions in a PCIe sled that is no larger than a standard GPU adapter. This would bring a cost-effective, NVMe storage solution in very high density to the market.
Big data is still growing and becoming increasingly more valuable. With the explosion of IoT devices and their uses, more data is being collected than ever before with the ability to garner insights that will lead to new business or revenue. The industry has been making leaps and bounds to increase density but have yet to deliver on the 1PB per 1U so far. While it is true one could take 128TB SSDs and put ten in one array, it would be extremely cost prohibitive. Dell EMC has come up with a new solution.
Dell EMC’s Extreme Scale Infrastructure (ESI) group is taking lower capacity and lower cost M.2 SSDs and putting 10 on a PCIe sled no larger than a standard GPU adapter. Currently M.2 capacity will deliver almost 100TB per sled with four sleds fitting in a Dell EMC C4140 placing it almost 400TB in a 1U form factor. Not quite the 1PB but M.2 drives will be more commonly found at higher capacities, most likely this year, bringing the total capacity up to nearly 800TB per 1U, much closer to the 1PB goal.
Dell EMC ESI is claiming that this can be done for SATA pricing, but these will be NVMe drives. This will be delivering NVMe performance for up to 800TB per rack space unit. To further optimize performance, these sleds (along with the PowerEdge C4140) will be paired with two 100Gb NICs for 200Gb bandwidth performance. So a 5U solution can deliver 1Tb of throughput and millions of IOPS, not to mention the storage potential.
This amount of storage and performance would be ideal for handling the sustained ingest of massive amounts of data, for example, as a front end in an edge computing architecture. This would also be good for companies leveraging Machine learning, Artificial Intelligence, or deploying large amounts of IoT devices.