System architects are under increased pressure to deliver scalable & predictable performance block storage services. In shared multi-tenant environments, the lack of predictable performance of traditional storage presents one of the biggest barriers to delivering business-critical applications from a shared infrastructure. To meet growing performance needs, administrators are often forced to use large cache layers, tiered disk systems or massive amounts of disks in large arrays.
The common result is marginal performance gain with a sprawling network of underutilized capacity that is impossible to manage and costly to operate at scale. In Short: Traditional disk-based storage architectures were not designed to accommodate the massive performance requirements of computing at large scale.