10K and 15K rpm HDD in their last gen, but WD releases mid capacity 3.5" for Big Data Analytics

[cbn]

https://www.forbes.com/sites/tomcou...eagate-products-at-the-2018-ces/#1290d54a2dc6

The hard disk drive companies (represented in this blog) I spoke with said that they had introduced their last generation of high-performance HDDs, these are 10,000 and 15,000 RPM HDDs used for enterprise applications. They all plan to only provide flash storage for high-performance applications, likely using the high-performance NVMe interface. In the enterprise space, capacity HDDs providing inexpensive storage will be the dominant application for HDDs.

Meanwhile Western Digital Is releasing some mid capacity 3.5" for Big Data Analytics:

http://www.cdrlabs.com/news/western-digital-expands-mid-range-enterprise-hard-drive-offerings.html

Western Digital leads the high-capacity enterprise hard drive market for hyperscale and cloud environments with its 10TB, 12TB and 14TB drives, where the company’s HelioSeal helium-sealing technology is required for the highest capacities and densities to deliver the lowest (or best) TCO. However, many applications have small data sets or benefit from optimal performance from implementation of a cluster of lower capacity, air-based hard drives. Many traditional data center systems still rely on moderate capacity points for RAID-based block-and-file systems.

Distributed file systems such as Apache Hadoop® often have workloads that require higher access density in order to maintain performance when running Big Data analytics jobs. The “sweet spot” capacities for these applications are 4TB, 6TB and 8TB. Western Digital’s new enterprise solutions are designed for this space, while providing enterprise-level performance and specifications for storage OEM partners to serve end customers with a range of capacity and price points while maintaining the enterprise integrity demanded by today’s data center architects.

So with SSD taking over the highest performance level, the balance of IOPS vs. capacity for HDD is shifting towards capacity but not completely. (re: Largest drives have reduce IOPs per TB density)

 

[Elixer]

I wonder how much will MAMR & HAMR tech along with adding more independent actuators change this picture?

Heck, I also wouldn't be shocked if they can come up with a MAMR based 15K RPM 20TB monster drive, and I am betting these beasts will sell like hot cakes.

 

[cbn]

I wonder how much will MAMR & HAMR tech along with adding more independent actuators change this picture?

That is exactly what I was thinking when I wrote this thread.

So assuming (for the sake of argument) areal density did not change a dual actuator drive would be needed at 8TB to 16TB capacity in order to duplicated WD's stated Sweet spot of 4TB to 8TB per single actuator drive for Hadoop.

But how much does this actuator per TB ratio change if areal density does increase? (eg, HAMR at 3TB and 4TB per platter) Maybe a greater number of actuators per drive (each of these being lighter and with a faster seek time)?