Re: [i2rs] Architecture Discussion 2: Persistence (ephemeral vs. permanent)

[Joe Marcus Clarke <[e-mail]>]

On 9/15/13 10:11 AM, Juergen Schoenwaelder wrote:
> On Sat, Sep 14, 2013 at 08:41:03PM +0100, Scott Whyte wrote:
>>
>> I'm not sure I agree that a client is a direct subscriber, at least in the
>> sense that brokers are very useful for scalability.  Using them I can
>> design a system where each agent talks to N brokers, where N is reasonable
>> for me (whether 2 or 20), and the broker-agent communication is independent
>> of the broker-client communication.  It might be very nice to have the
>> broker-agent communication be UDP and the broker-client communication be
>> TCP for example, where I purposefully put the broker topologically close to
>> the agents and the clients can be hierarchically distributed throughout the
>> network.
>>
> 
> I am not sure what exactly a 'broker' is. I did not see this term
> defined in the architecture document nor in the problem statement
> document. Whatever this 'broker' is precisely, is it needed for I2RS
> 1.0 or something that may come along later?

It was not described in the architecture, but it is not an entirely
foreign concept to I2RS discussions.  Nancy Cam-Winget brought it up
during the IETF-87 session, and it is described in
draft-camwinget-i2rs-pubsub-sec-00.  I realize this is not a WG document
yet, and that the broker is not an element of the architecture yet, but
a broker does represent a way of doing scalable pub-sub.

Joe

-- 
Joe Marcus Clarke, CCIE #5384,         |          |
SCJP, SCSA, SCNA, SCSECA, VCP        |||||      |||||
Distinguished Services Engineer ..:|||||||||::|||||||||:..
Phone: [phone]         c i s c o  S y s t e m s
Email: [e-mail]

----------------------------------------------------------------------------