4 ms·
How SSDs conquered mobile devices and modern OSes
- PakG1 14y agoIt's interesting how this has played out exactly the way Clayton M Christensen predicted in his Innovator's Dilemma book years ago. Back then, flash was barely on the radar of most consumers, and yet, Christensen identified flash as a primary candidate to make traditional harddrives obsolete the same way desktop harddrives had made mainframe harddrives and minicomputer harddrives obsolete. Chalk one up for the guy, his Innovator's Dilemma model is really well thought out.
- ajross 14y agoIs this really so surprising? I think anyone who knows how semiconductor manufacturing works would have looked at the scaling graphs of magnetic storage vs. transistor density and seen that they cross. We've all been using flash-based embedded systems for well over a decade -- it started with cameras, then fixed-function network gadgets, then phones, then... So now it's hitting PCs. If anything, magnetic storage has actually scaled faster than expected (frankly it's so cheap in bulk that it's effectively free relative to the power required to run it). To me, the only surprise was that consumer applications never found a need for more than about a quarter TB of storage, so that's where the crossover happens on PCs. We haven't actually hit the cost-per-byte point yet.
- edge17 14y agojust a datapoint about 'obsolete,' tape is still used for backups and remains attractive in terms of cost and storage density
- hobbyist 14y agoHere is some neat work done for using SSD's effectively in network appliances and data cache servers. http://www.cs.princeton.edu/~abadam/ssdalloc.html http://www.cs.princeton.edu/~abadam/ssdalloc.html
- jrabone 14y agoIt's just as well that 256GB is enough for anyone. Maybe I'm pre-empting the next part in this series, but it looks like simply scaling NAND flash geometry might be a dead end: "While the growing capacity of SSDs and high IOP rates will make them attractive in many applications, the reduction in performance that is necessary to increase capacity while keeping costs in check may make it dif- ficult for SSDs to scale as a viable technology for some applications." http://cseweb.ucsd.edu/users/swanson/papers/FAST2012BleakFlash.pdf http://cseweb.ucsd.edu/users/swanson/papers/FAST2012BleakFla...
- freehunter 14y agoWhile SSDs are still scaling, like you said they won't scale forever. Memristors have just recently been proven to exist and there are a lot of companies showing that they have working memristors. I hope that by the time we reach "peak silicon" on SSDs, memristors will be there to pick up the slack. They've been a long time coming, but now that they finally exist we should be seeing them soon. According to HP, they should have memristors on the market in about 6 months. http://www.dailytech.com/HP+to+Deploy+Memristor+Powered+SSD+Replacement+Within+18+Months/article22963.htm http://www.dailytech.com/HP+to+Deploy+Memristor+Powered+SSD+...
- miahi 14y agoWe will probably use a combination of SSDs for speed and HDDs for storage for a long time. I don't think it's cost effective to keep everything on a SSD (like: movies), as the cost/GB is ~15x higher than the HDD storage. I have a combination of SSD+HDD in both my machines (desktop: first-generation 40GB SSD + 2TB 7200rpm HDD, laptop: 128GB SSD + 500GB 7200rpm HDD) and I get the best from both worlds: cheap storage for large files on HDDs and fast OS and applications on SSDs. Because of this combination I don't feel the need to use a bigger (and costly) SSD.