"Generally 2.5-inch disk drives have higher areal densities than 2.5-inch drives "
Seagate has demonstrated heat-assisted magnetic recording technology with 1 trillion bits per square inch, a 30 per cent improvement in Toshiba's production record of 744Gbit/in2 with its MQ01ABD drive – a 2.5-inch, 5400rpm, SATA II drive. Seagate's highest production areal density is thought to be the 3TB Barracuda, a 3- …
Surely there would in fact be an improvment in IO. When spinning at similar RPM's the read head should pass 20x more data at any given location on the disk purely because the density is increased.
It happen with PMR, transfer rates went up significantly. I think that it wont scale exactly to match the increased density but there should be some improvement.
Transfer rates (for sequential IO) will surely go up.
Access time, aka seek time, or latency for anything more random will go up or stay the same.
Hopefully flash based cache of ~4-32GB range will help solve some of this, but it's still going to be there for non-sequential IO.
I agree. If the densities go up as much as they say, the Data Transfer Rates are going to go up substantially.
I did a lot of disk benchmarking in the last job. And sure enough, the DTR did go UP with the higher data densities of PMR drives. In fact the higher densities resulted in 5400rpm drives with a better DTR than the prior gen 7200's. The 7200's bumped up to the previous 10K DTR, the 10K's ended up getting close to the 15K's, while the 15K's ended up being nearly the same as before (never got an explanation for that). So our contract ended up spec'ing 7200 and 10K drives and saved money by not buying the 15K's on the higher end that time.
For myself personally, I ended up going with new 5400rpm drives on a couple of laptops to replace old 7200's that were needing replacing anyway and saved some on power. Performance still went UP.
Benchmarks are a good thing. :) You can't trust plain RPM numbers these days due to differing data densities. You actually have to verify that a 7200rpm drive is faster than the 5400, or that the 10Krpm is actually faster than the 7200.
Now, what I did find was latency was where the higher rotation speeds still made a difference on the newer drives. The higher RPM's still resulted in lower latency scores, but once the data file was locked on, the transfer rate was higher than I expected. That's in a perfect world of fresh benchmarking. Now why would you need lower latency - Disk fragmentation would be a big one. :)
A 20x increase in areal density would equate to increased data rate of 4-5x (or root of 20), and a similar increase in the number of tracks per platter.
Anyway, this article makes no mention of the 'salting' discovery reported recently, which ought to provide a cheap way of boosting the capacity of current PMR drives by a few multiples.
It will be interesting to the see the impact of frickin' laser beams inside the drive upon the heat output of these drives. Given that our current disk failure rate is already around 1/fortnight in our three-initial vendor arrays, I expect we'll have to permanently station someone in the datacenter just to swap drives.
"Seagate suggests initial HAMR technology could provide 6TB for 3.5-inch drives and 2TB for 2.5-inch models"
Meanwhile, an article with the same date as this one:
WD intros 'world's first' 2TB portable HDD
"They're available now from WD's online shop and its many retailers"
Did they just sneak more platters in??