Silent_Buddha
Legend
That's interesting, but not entirely applicable to consumer devices going into the future.
Unless something changes, performance HDD's are going to be stuck as SMR (the current future of expanding bytes per inch on platters) is the way forward unless you get into far more expensive Helium filled enclosures (and even these have a limit). While SMR is excellent for data centers that can work around its critical weakness (incredibly bad write performance) consumer devices aren't going to fare as well. The Google white-paper goes into a little bit of detail on the challenges of using SMR based drives for a data center. Problems that can't be solved easily or cost effectively on consumer devices.
Seagate has released some SMR drives into the market with impressive 1.33 TB/platter densities. However, it also needed to include 20 GB (GB not MB) of cache in order to make common tasks useable. Once you exceed that cache, however, write speeds drop to ~10 MB/s or less (not uncommon for modern games to exceed 20 GB). They are fantastic for data centers which archive lots of data as read speeds remain decent and high capacity drives based on SMR are significantly cheaper than high capacity drives based on PMR (the current performant platter tech for high capacity).
So yes, in enterprise data center's SMR has extended the period of time before SSD's can replace most data storage installations. But unless something else comes around HDD capacity increases are going to slow drastically. There's hope with HAMR based platters, but that has been plagued with issues and potential launch dates keep getting pushed back (2018 is the latest optimistic projection), however, there are people within the industry that doubt HAMR will ever be a commercially viable technology.
If HAMR can become commercially viable and if it can be done more cheaply than NAND based devices (no guarantee) then physical HDDs can remain relevant for a very long time. If not, NAND (or other solid state) devices will converge with physical platter based devices sooner rather than later.
Regards,
SB
Unless something changes, performance HDD's are going to be stuck as SMR (the current future of expanding bytes per inch on platters) is the way forward unless you get into far more expensive Helium filled enclosures (and even these have a limit). While SMR is excellent for data centers that can work around its critical weakness (incredibly bad write performance) consumer devices aren't going to fare as well. The Google white-paper goes into a little bit of detail on the challenges of using SMR based drives for a data center. Problems that can't be solved easily or cost effectively on consumer devices.
Seagate has released some SMR drives into the market with impressive 1.33 TB/platter densities. However, it also needed to include 20 GB (GB not MB) of cache in order to make common tasks useable. Once you exceed that cache, however, write speeds drop to ~10 MB/s or less (not uncommon for modern games to exceed 20 GB). They are fantastic for data centers which archive lots of data as read speeds remain decent and high capacity drives based on SMR are significantly cheaper than high capacity drives based on PMR (the current performant platter tech for high capacity).
So yes, in enterprise data center's SMR has extended the period of time before SSD's can replace most data storage installations. But unless something else comes around HDD capacity increases are going to slow drastically. There's hope with HAMR based platters, but that has been plagued with issues and potential launch dates keep getting pushed back (2018 is the latest optimistic projection), however, there are people within the industry that doubt HAMR will ever be a commercially viable technology.
If HAMR can become commercially viable and if it can be done more cheaply than NAND based devices (no guarantee) then physical HDDs can remain relevant for a very long time. If not, NAND (or other solid state) devices will converge with physical platter based devices sooner rather than later.
Regards,
SB
Optical media will be following in data cassette's footsteps. It's just a question of when.
