I spy a possible approaching tipping point.
Could greening the datacentre threaten disk? Is this remotely likely? Consider this: IBM has just announced a blade server with a solid state disk in the space previously used by a 2.5 inch drive.
So why am I getting excited?
The disk has 15.8GB capacity, enough to enable either 32- or 64-bit Windows to boot from it. It comes in a single or dual format, meaning a second 15.8GB holding a copy of the first 15.8GB's contents for safety. Alternatively you could use all the 31.6GB capacity with no RAIDing. VMware and Unix support is probably coming. It costs $499 for a single SSD and $999 for the dual one.
What you could do is replace a 73GB 2.5inch SAS drive on the blade with this for the O/S boot function. The boot would take a few seconds longer so why would you do this? It's because so much less electricity is needed.
IBM's 15.8GB SSD is 95 percent more energy efficient than a 73GB SAS drive, according to IBM. In fact, "These 15.8GB drives use only 2W per drive versus as much as 10W per 2.5 inch hard drive and 16W per 3.5 inch hard drive. The resulting power savings can be up to 1,512 watts per server rack, with 50 percent heat reduction.
Flash drive prices are said to be falling 50 percent a year with capacity doubling annually and the trend continuing through 2009. Let's extrapolate some numbers:-
2007 15.8GB $499
2008 15.8GB $250 31GB $499
2009 15.8GB $125 31GB $250 62GB $499
2010 15.8GB $63 31GB $125 62GB $250 124GB $499
Today I could buy a Fujitsu 2.5inch 73GB SAS drive for $299.
Although a $200 premium for less than a quarter of the capacity is a hefty cost it is not totally outrageous, particularly not when energy costs are brought into play.
Electricity costs are rising. At some stage the working lifetime cost of ownership of hard drives, taking into account energy costs, will exceed that of equivalent capacity flash drives.
SSDs are more reliable than hard drives. IBM reckons that in large installations 3 percent of hard drives fail a year versus 1 percent of SSDs. It's used wear-leveling algorithms to get over the short life cycle of flash.
SAS drives tend to perform faster than flash SSDs for random and sequential write workloads and are slightly faster than a flash-based SSD for Windows boot. Flash excels in random reads.
If flash-SSD costs reduce faster than hard drive costs and if energy costs continue to rise and if flash SSD lifetime can be increased still further then maybe flash SSDs will start to eat into 2.5 inch hard drive sales. At present 2.5 inch flash SSDs will only go where hard drives aren't suitable, telephone exchange applications perhaps and high vibration/extreme temperature environments.
But green concerns could tip the hard drive/flash SSD balance in favour of flash a little more strongly year by year. Combine that with flash SSD cost reduction and capacity increase and a real tipping point might be approaching.