Interesting article, thanks!
I still think the flash/HDD transition point for bulk storage is quite a few years away. I'm a huge fan of SSDs and was a relatively early adopter. I paid around $350 Australian way back in 2009 for a 64GB SSD, didn't regret it once! (It's still driving the HTPC under my bed, by the way, hasn't missed a beat, and it ensures the system boots almost as quickly as the TV turns on).
But today (or in the next few years), ditching HDDs completely? There's some problems I think with that approach.
If we look at pricing today:
-> If you only ever need ~500GB or less - for sure go all flash, no question there.
-> If you only ever need ~1TB or less - SSDs are still around the ~$170 mark, and that's generally for the bargain basement SSDs. You'd save money on a higher quality ~250GB SSD and a 1TB HDD. Having said that, you could absolutely make a good case for a single 1TB SSD.
-> As soon as you need 2TB or more though, the price of the all-flash solutions start looking pretty terrible.
Let's suppose SSD prices halve, at that point (whenever it is) 2TB all flash solutions might start making sense, but for anything 4TB or more, all-flash solutions will still be vastly more expensive than an SSD + HDD mix. In other words, even if SSDs halve in price, 4TB and larger HDDs will still play an important role in the market. Even when SSDs drop to a quarter of current pricing, an 8TB and larger HDD will likely still be vastly more cost-effective than going all-flash storage.
I also have my suspicions with the proposed "solution" of a single, large, budget SSDs, over the current SSD + HDD combination that so many of us currently rely on. There's very little perceived difference between mid-range and high end SSDs even with most intensive desktop workloads these days. HOWEVER, the performance gap between the budget and mid-range drives is noticeable and I think there's reasonable signs that the gap will grow in coming years. We're going to see budget drives increasingly ditch on-board RAM, move to QLC and use cheaper/simpler controllers with fewer channels and slower response times. Intel's new 660p gives us a good example. Once the write buffer is full the sustained writes to the drive drop to ~100MBps. That's slower than a cheap laptop drive and could rear it's head doing something like copying video footage off a camera over USB3. Other drives which rely on system memory and/or ultra-budget controllers show different, but also tangible, performance limitations.
To be clear, I have no issue with these budget drives existing on the market. The point is that there are genuine performance penalties in driving SSD costs down. I also think that anyone profiting from selling SSDs and SSD components are going to want clear market segmentation between budget and mid-range drives. They want those of us who care about performance to have to spend more money on a mid-range drive. That happens less if budget drives become fast enough for the needs of enthusiasts (whatever "enthusiast" really means!?).
So thinking about replacing the SSD + HDD mix with a single SSD: any setup that is even remotely cost-effective generally requires you to look at the cheapest of the cheap drives and use a single drive for the entire system. But then your system is booting and running applications off one of those ultra-budget drives. Do we really want that?
Sure, you could stay with a two-drive solution and have a lower capacity but higher end SSD for system + programs and a cheap SSD for bulk storage, but I don't think that makes sense in most cases either. It costs more, for a start. You're still stuck with the hassle of managing multiple drives (like a standard SSD + HDD setup). Finally, even if you were prepared to pay more for this sort of setup, you might find you'd be better off swapping out the large & (relatively) cheap SSD for a HDD and use the (significant!) money you saved getting a higher capacity mid-range SSD for the system, programs and games, with the HDD for things like videos and photos which rarely benefit from flash storage anyway. The author has identified an interesting benefit for a photo library on an SSD - generating thumbnails for a folder with thousands of photos - that makes sense, but I sort my photos so I've never run into that issue. I suspect it's a bit of an edge-case. In my experience, once the photo, audio or video editing is done and those final files are simply sitting there waiting to be viewed or listened to, there is very little if any benefit to those files sitting on flash rather than spinning metal storage.
TLDR:
A single, large,
mid-range SSD is going to be
very expensive into the future
A single, large,
budget SSD is going to perform
worse in some situations over the common mid-range SSD + HDD combo
A small, mid range SSD and large budget SSD is also more expensive and IMHO doesn't make sense over the common SSD + HDD combo in lots of circumstances anyway.