Seagate hits 1Tb/square inch areal density

[BFG10K]

Heat assisted magnetic recording (HAMR): http://www.seagate.com/ww/v/index.j...toid=295d922d58716310VgnVCM1000001a48090aRCRD

The technology offers a scale of capacity growth never before possible, with a theoretical areal density limit ranging from 5 to 10 terabits per square inch – 30TB to 60TB for 3.5-inch drives and 10TB to 20TB for 2.5-inch drives.

 

[FAUguy]

I read about that a few hours ago. 30-60TB HDD sounds nice, but if the drive is going to only have a 1-2 year warranty and cost a lot, I doubt many home consumers will buy it. Though may be targeted more for enterprise and home servers.

 

[Anteaus]

Reliability is always a question when brand new techniques are introduced, but its a great thing if it works. I guess we'll have to wait until some drives ship and see what happens. With SSD rapidly gaining mainstream acceptance, these will likely be predominately backup or media drives. I'm excited about it.

My biggest concern is not whether it works, but what type of error rates we are going to get as things get smaller. I bet these things are going to need some serious ECC.

 

[groberts101]

nice to see that the mfgrs are testing the waters with this tech. We need to crawl before we can run and eventually they'll get it sorted if they throw enough cash at the process.

I believe they are facing the simple fact that SSD is here to stay and only gaining momentum/market share at this point, so they must invest in whatever way possible to keep things going. I rarely worry about initial pricing since they will surely refine the processes and mfgr techniques, not to mention that demand will go up as data centers adopt it. Then comes the competition and mainstream adoption years later.

Which is fine by me since it's far better to move ahead slowly.. than not at all when it comes to mass storage. Regardless of all the hypothesis out there.. it will likley be many years before we can buy a 4TB SSD for $200. And that being said.. I'll gladly save me pennies to buy an ultra-fast 10TB HDD even if it costs a $1000.

 

[exdeath]

But we would *still*, all these decades later, only access it at a whopping 100 MB/s sequential and < 1 MB/sec random. Anyone care to do the math on time wasted filling and emptying 60 TB at 1 MB/sec?

No thanks.

 

[groberts101]

But we would *still*, all these decades later, only access it at a whopping 100 MB/s sequential and < 1 MB/sec random. Anyone care to do the math on time wasted filling and emptying 60 TB at 1 MB/sec?

No thanks.

umm.. where are you getting that info?.. cause I think you may need to educate yourself with this tech before assuming those kinds of speeds.

Especially since this tech is said to have the potential for terabytes of information per second.
http://www.theregister.co.uk/2012/02/08/boffins_laser_storage/

 

[ArchAngel777]

But we would *still*, all these decades later, only access it at a whopping 100 MB/s sequential and < 1 MB/sec random. Anyone care to do the math on time wasted filling and emptying 60 TB at 1 MB/sec?

No thanks.

As platter density increases, so does the performance, assuming the same rotational speeds. Seek time and access time will probably remain the same, but because more data is stored per square inch accessing it will be quicker overall.

That is generally why people [enthusiests] seek out to know the platter size of a hard drive when making a purchasing decision. Generally speaking, a 3 platter 1TB drive will outperform a 5 platter 1TB drive. Although, I am not sure if anyone even makes a 5 platter drive anymore. Seems reliability is an issue with that.

This is really good news for mechanical storage. SSDs have been gaining ground, but lets face it, mechanical storage is still the only realistic option for, well, 'storage'.

 

[Elixer]

Unless those drives are cheap enough to purchase in a two pack, that is just too much data sitting there waiting for something bad to happen...

 

[exdeath]

As platter density increases, so does the performance, assuming the same rotational speeds. Seek time and access time will probably remain the same, but because more data is stored per square inch accessing it will be quicker overall.

That is generally why people [enthusiests] seek out to know the platter size of a hard drive when making a purchasing decision. Generally speaking, a 3 platter 1TB drive will outperform a 5 platter 1TB drive. Although, I am not sure if anyone even makes a 5 platter drive anymore. Seems reliability is an issue with that.

This is really good news for mechanical storage. SSDs have been gaining ground, but lets face it, mechanical storage is still the only realistic option for, well, 'storage'.

In the time hard drives have gone from 1GB to 4000GB, transfers rates have only gone from 40MB/sec to not even quite 150MB/s and random access times have not changed since voice coils replaced stepper motors some decades ago.

Moving 4000 times the data at 4 times the speed equates to being 1000 times slower. Contrast to SSD, performance scales with size; double the NAND channels gives double the capacity, double the iops, double the transfer rate.

 

[frostedflakes]

What 1GB drives could do 40MB/sec? ST51080A, for example, had an interface speed of only 16.6MB/s according to the spec sheet and actual transfer rates were probably lower (similar to how modern HDDs can't even saturate SATA 3Gbps, much less SATA 6Gbps).

http://www.seagate.com/support/disc/manuals/ata/51270c1.pdf

Also some newer drives, such as the 1TB/platter Seagates, can get closer to 190-200MB/s sequential transfer rates on the outside of the platter

Transfer rates still don't seem to have scaled linearly with areal density and capacity, but the scaling isn't quite as bad as you make it out to be.

 

[Anteaus]

As it has been said, this really isn't an either/or situation. SSDs are going to be established as primary use drives as prices drop. They are already more than large enough to handle day to day usage and the performance characteristics are far beyond hard drives. Assuming they can figure out how to deal with the long term write degradation, this is inevitable.

Next generation hard drives are going to be primarily storage devices. Even so, it would be naive to think there won't be advances that will enhance their performance.

I'm curious as to if we may see SSD hybrids, where SSD and harddrives are combined into one unit. It wouldn't be hard for a drive to determine which files are read more often and stores them in the SSD portion, whereas infrequently used files are relagated to the hard disk. Also, since the interface is internal transfer speeds would be much faster than through an independent controller. This is just a thought of mine so if they do exist in real life even as test beds it's completely coincidental.

 

[Smartazz]

It's interesting that hard drives continue to increase in density at roughly an exponential rate like integrated circuits while being manufactured significantly differently.