RE: Requirements specification

[Stephen Byan <Stephen.Byan@quantum.com>]

Steph wrote:

> We concluded that as long as channel speeds were increasing briskly it
> didn't make sense to spend any effort on a low level striping scheme.
> If network technology generations start coming years apart, it might
> make sense again.

I don't think network channel speeds are increasing briskly. Ethernet was 10
Mb/s in 1980, 100 Mb/s in 1990, and 1 Gb/s in 2000. These are slow
infrastructural step functions. 

Note that continued increase in storage capacity is based partly on a
continued increase in media bandwidth. Also part of the modest increase in
random access performance over the past decade has been based on a continued
increase in media bandwidth (i.e. higher spindle RPM). 

In the 1990's there was a transition from shared bandwidth to switched
bandwidth which helped mitigate the problem of the slow evolution of link
speeds. I think storage interconnect bandwidth needs to be able to scale
nicely between the step-function increases provided by new physical layers.
We've already spent the scaling available by moving from shared bandwidth to
switched bandwidth; link aggregation seems like the next necessary step. 

I argue that for storage it's not enough to aggregate links statistically,
based on multiple flows. Instead it is necessary to aggregate links for a
single flow. Consequently I now favor the iSCSI session concept.

Regards,
-Steve

Steve Byan
<stephen.byan@quantum.com>
Design Engineer
MS 1-3/E23
333 South Street
Shrewsbury, MA 01545
(508)770-3414
fax: (508)770-2604