Re: [Anima-bootstrap] IoT and scope of bootstrap

Hi Brian,

Brian E Carpenter schreef op 2015-12-01 20:11:
> Hi Peter,
> 
> On 02/12/2015 02:11, peter van der Stok wrote:
> ...
>> Such a network will be composed of LLNs, and includes constrained 
>> devices.
>> Sleepy nodes (energy harvesting sensors) may include ANIMA services, 
>> but special provisions to include them in the network are
>> more likely to be deployed.
>> 
>> For the moment I have no idea what ACP and Anima bootstrapping implies 
>> in terms of resources and resource consumption.
>> But excluding everything smaller than a tablet may be premature.
> 
> ANs are basically nodes that participate in management actions that 
> historically
> would have been carried out by a centralised NMS or NOC. So in my view
> an individual
> light switch isn't an AN but should be managed by an AN. The lighting 
> controller
> for a section of a building could be an AN. Or to say it another way,
> a node that
> just does what it's told isn't an AN.
<pvds> Strictly speaking from a network management point of view?
In that case the numerous light controllers are probably not ANs.
However, the building control can (will) be part other networks, and the 
management of all these network components will include ANs.
For the moment these aspects are far from clear to me.
</pvds>
(It may of course need a
> security bootstrap
> anyway.)
<pvds>
That's where my interest comes from. Listening to Max in Yokohama, where 
he explained his central table, I understood the bootstrap
and the accompanying ACP to realize a domain wide neighbor discovery 
including secure bootstrap (beyond single link or mesh); with apologies 
for my over-simplification.
The domain being (one of) the building control component(s) in the total 
network.
For the moment I did not see the need for a multi-threaded OS to realize 
this part.
Therefore my optimism that (this part of) Anima could be used to 
separate network installation responsibility from application 
installation responsibility.
</pvds>
> 
> However, indeed the resource question is important. An AN is going to 
> need a
> multi-threading OS, a full network stack, and a heap of software on 
> top. For
> example, GRASP alone is about 2500 lines of C (in the now-obsolete
> BUPT prototype)
> and looks as if it will also be a couple of thousand lines of Python. 
> Add to
> that the ACP, the security bootstrap, and all the supporting code, 
> before
> you install any ASAs.
<pvds>
Therefore I will try to look for possibilities to install ACP with 
bootstrap in a given node without using all the other goodies.
Unless such a separation is unwanted, and my view needs to find 
expression elsewhere.
</pvds>
> 
>     Brian
> 
> _______________________________________________
> Anima-bootstrap mailing list
> Anima-bootstrap@ietf.org
> https://www.ietf.org/mailman/listinfo/anima-bootstrap