Seagate 15K.2 Inflection Point - The New Era of Serial Attached SCSI

Introduction

Necessity can indeed be the mother of invention, and over 20

years ago it gave birth to an innovative interface that went on

to achieve worldwide success: parallel SCSI. To be sure, the

authors of the original parallel SCSI standards well understood

the theoretical superiority of serial interfaces, with their inherently

simpler and more robust architecture. Unfortunately, serial

technologies at the time were woefully slow, adequate for the

pedestrian needs of peripheral devices (for example, keyboards

and mice) but far too sluggish to efficiently move the multi-

megabyte files increasingly found on primary storage devices.

And so parallel SCSI (and its desktop counterpart, parallel

ATA) came into being; not because parallel interfaces are

fundamentally preferable (quite the contrary), but simply because

they were the best option at the time, given the limitations of

available technology. The premise, and promise, of parallel

architecture was certainly compelling: to bypass the performance

bottleneck of a single signal path (sending bytes

serially

, or one

bit after another), just add more signal conductors to enable

a byte’s multiple bits to be sent concurrently (or in

parallel

),

each on its own separate data path. The simple logic seemed

unassailable—an eight-lane highway can certainly move far more

traffic than a single-lane one.

To be sure, practical implementation of parallel interface theory

proved decidedly complex, with numerous technological hurdles

to overcome. Nevertheless, the fact remains that for its time and

purpose, parallel SCSI was a profoundly important advancement

in storage interface technology.

But, of course, times change.

SCSI Inflection Point:

Standardizing on

Serial Attached SCSI (SAS)

Technology Paper