[FoRK] PubSub NG Re: MQTT : Exploring the Protocols of IoT - News - SparkFun Electronics

Gregory Alan Bolcer greg at bolcer.org
Fri Feb 27 14:11:45 PST 2015


Cloud spending is predictable and expensive.

I'm guessing Fog computing is unreliable, cheap, and scary.
http://www.imdb.com/title/tt0080749/

Greg

On 2/27/2015 1:44 PM, J. Andrew Rogers wrote:
>
>> On Feb 27, 2015, at 11:58 AM, Ken Meltsner <meltsner at alum.mit.edu> wrote:
>>
>> Cisco is apparently pushing a new buzz phrase of "fog computing" --
>> just like "cloud" but distributed to the network's edge.  Anything
>> that fits that nebulous description will probably get lots and lots of
>> attention.  Support for mobile code?  Results aggregation in a mesh
>> network?  Mix and match your favorite buzzwords....
>
>
> Cisco is a buzzword factory. I do not think “fog computing” is a great term but it reflects something real.
>
> For a machine-generated sensor and telemetry data, there is not enough bandwidth to backhaul all the data because aggregate bandwidth is a function of distance. Additionally, many of the platforms are fairly specialized for the sake of efficiency. Because the analytics are frequently operational, this is creating the requirement for a compute topology that approximately matches geophysical topology. To make matters worse, many platforms are mobile which means the resource topology is not fixed. People are already working on autonomous compute-optimized sensor-less motherships that can locally coordinate sensor platforms that lack sufficient compute on their own.
>
> This creates a need for constructing temporary ad hoc computing fabrics out of remote resources that in aggregate have sufficient sensors and resources to get the job done. Our software infrastructure is extremely not ready for these operational models but they are coming anyway.
>
> Futurists talked about this for decades but it is about to become reality.
>
>
> _______________________________________________
> FoRK mailing list
> http://xent.com/mailman/listinfo/fork
>

-- 
greg at bolcer.org, http://bolcer.org, c: +1.714.928.5476


More information about the FoRK mailing list