SSD's make terrific sense for typical VPS users and I would be willing to pay more. I might move if I can't get an SSD timeline. Here's why:
Database is typically the limiter for most web applications. The primary reason for using a VPS compared to a more managed host is to run web applications that need databases. Although I can optimize MY disks to write mostly sequentially and buffer MY logging to happen in chunks and optimize MY daemons to not log excessively, the OS has to access disks randomly to serve all neighbors and they might be consuming lots of IO that I can do nothing about.
VPS's, because of the presence of many applications on one machine, are by their very nature random-access I/O. SSD's make excellent sense for improving the performance of the rest of the machine. I can burn IO in incredible amounts by utilizing tons of async server optimization and limiting database fsyncs, and I'm sure there is some concern that SSD clients will all be better at being bad neighbors, but I'm willing to bet there are just as many clients that want the SSD performance boost instead of investing in the skill to totally preoccupy the disk I/O as much as possible.
Quote:
jGreat! The host OS gave me the keys! Now let me wait 27,000,000 cycles so the disk can get back to the WAL and my WSGI process can finish a huge pile of transactions... - my kernel
The limiting factor on all dynamic websites is eventually the database. Async, optimized fsync,
and caching can only go so far. I would much rather have an SSD with 10x the random IOPS than pay 8x more for 8x the memory, 8x the CPU, and 8x the IO tokens. On the face of it, it makes cost sense to customers to upgrade the bottleneck instead of paying for a bunch of heat.