[1st-mile-nm] [FRIAM] Kicking the Internet up a notch

[Bob Knight]

I agree about the "catch up". In my "day job", I have serious concerns 
about the way we're delivering computing from any number of directions: 
costs, resource (e.g. people) utilization, encryption (whole-disk and 
partial), security, compliance/accreditation etc. whilst keeping 
enterprise management in mind. While not suitable for some situations, 
the Sun Rays (as an example) provide solutions or easily tractable 
workarounds  to many of the aforementioned problems. They also drive 
some costs to service (e.g. bandwidth) providers, which is just fine by 
me.From a field support perspective, for example, I sure like the "lick 
and stick" model when one (rarely) keels over. In an era of "doing more 
with less", there's a lot to like. I anticipate, for instance, that I'll 
have a SR on my desk Real Soon Now, with a SR laptop for working at home.

And I'm reminded of the similarity of centralization to running 
DECsystem-20's in the 80's at NMT, Stanford and SRI...perhaps I'll 
resurrect my white lab coat :).

Bob

John Osmon wrote:
> On Sun, Apr 06, 2008 at 09:16:00PM -0600, Bob Knight wrote:
>   
>> Am I missing something or does this dovetail nicely with something like 
>> LTSP or Sun Rays?
>>     
> [...]
>   
>> Or is this completely tangential?
>>     
>
> Nope -- things are just starting to catch up to the promise we've had
> for a while...  
>
> In the mid-90s, I helped transform a single unix minicomputer into a lab
> of 50+ workstations that were interchangable.  Sit down at any, and your
> environment would follow (once you logged in).  Centralized storage,
> interchangeable displays, and a network were the key bits.
>
> The Sun Rays took that a bit further -- but few people were really
> ready for them...
>
> A friend was in town last week talking about a reseach project where
> the I/O bus of a computer was replaced with TCP/IP.  With the
> right bandwidth/latency, you don't *care* about separating the
> I/O and computing equipment.
>
> Back to 1st-mile issues -- thin clients have always been useful
> where there is sufficient bandwidth.  As we get better connections,
> the distinction between where computing components "live" will
> blur.  
>