Because physical limitations keep even the fastest mechanical disks performing almost a million times slower than RAM (Stockdale, 2007) for accessing random data it is unlikely that they will ever cease adversely impacting computer performance if used for that function. No matter how fast they spin or how small they get magnetic disks feature the attributes of physical size and motion that assure they cannot seek as quickly as electronic storage. That being said, it is notable that the throughput of sequential reads and writes is roughly comparable between the media (Murphy, 2008) and this may point to how they can continue to be useful in the future.

Radial size of the platters is very much a factor in seek performance (Chriswell, 2008). Not only can a smaller platter spin faster without disintegrating but there is less distance for the heads to travel when changing cylinders. So as miniaturization allows drive designers to fit more data into a smaller physical area we can expect commensurate latency and seek-time improvements (Mellor, 2009).

I have often wondered why there aren't more fixed head disk designs (, 2010). Obviously there must be some untenable economic factor but whether that is expense of manufacture, weight or some other aspect I've never found well-documented. The obvious gain of zero-seek time seems to me a benefit worth some trade-offs. As a long-time user of multi-track tape for both musical and data storage applications it has always seemed a straightforward solution to me. But Solid State Disks will shortly have rendered them moot anyway.

Increased disk cache size is another simple performance enhancement, one that has been very widely adopted. However this is really only a massive gain for sequential reading or writing, so its benefit is most apparent in the above mentioned throughput. While smart caching improves random seek times (Mehdi, 2009) the electronic memory to eliminate them would have to be such a significant proportion of disk capacity the expense would be roughly equivalent to a SSD.

Which brings us to the probable real solution to non-volatile random-access storage: The Solid State Disk (Murphy). With no moving parts they have negligible seek times and latency. Size is unimportant and cache is almost pointless. There will be improvements in throughput that may involve hardware or software caching and the maximum write/erase cycle (Mehling, 2009) issue must continue to be addressed but it seems likely that the era of mechanical disk drives as on-line storage will come to an end in the foreseeable future. However their price and capacities will keep them viable as near line storage for some time.

Chriswell, Chad (2008) SATA Hard Drive Performance Tutorial [Online]. Available from:

(Accessed: 23, January 2010).

Mehdi, Bilal (2009) 'Hard Disk Drive Specifications Guide: What to look for when buying disk storage', [Online]. Available from:,289142,sid181_gci1364141,00.html (Accessed: 23, January 2010).

Mehling, Herman (2009) 'SSD Makers Wrestle With Performance Degradation', [Online]. Available from: (Accessed: 23, January 2010).

Mellor, Chris (2009) 'Disk Platter Sizes Vary With Drive Speeds', The Register [Online]. Available from: (Accessed: 23, January 2010).

Murphy, David (2008) 'The (new) fastest hard drive ever', Maximum PC [Online]. Available from

(Accessed: 23, January 2010).

NA (2010) 'Fixed Head Disk Definition', [Online]. Available from:,2542,t=fixed+head+disk&i=43236,00.asp (Accessed: 23, January 2010).

Stockdale, Terry (2007) 'Speed of Virtual Memory Versus RAM', Terry's Computer Tips [Online]. Available from: (Accessed: 23, January 2010).