Fujitsu hits the data accelerator and parallelises SSD access

I've found that a sizable number of RAID 0 SSD's can have a significant grade of improvement even without direct addressing the Flash chips. Very data layout dependent though so you have to work it out. Nice to see someone else looking at it, especially at the chip layer.

I'll probably die before seeing it but this tied to in-memory storage would definitively set a performance bar, much in a similar performance jump is being seen with spinning disk -> SSD -> memory does for now. All that battery backed memory code becomes suddenly relevant again. (Hey, IBM, is this part of your storage business plan? Just askin.)

Yep, pre-beating my horse again.