Yet another report on RDBMS on HDD vs. SSD passes my bleary eyes. It's not materially different from all the others I've seen. And that's still a problem.
The whole point of SSD storage (or fully in-memory, a la HANA) is that Organic Normal Form™ schemas minimize storage footprint, maximize DRI, and are implemented through random I/O. Large RAID10 arrays of very fast fat HDD have been around for a long time, so SSD fast random I/O has been available longer than SSD, just not as cheaply.
The only meaningful test of RDBMS on HDD vs. SSD requires fully refactoring a lousy RBAR schema on HDD (the kind that "does transaction control in the client because that's what we're used to") to Organic Normal Form™ on SSD. Nobody bothers, that I've been able to find. It's too bad, because such tests end up being little more than rescaling of their respective sequential performance. And, yeah, SSD is faster. I think we knew that.