[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: conceptual questions



hi
	I would try and answer some of the questions and the rest can be
answered by the gurus along with any corrections in mine.


> (a) ...packets arriving from all links end up into the node entry?

Yes. They do. Even packets from the agents of that node end up at the
node entry.


> (b) ...that there is no queue between the node entry and the address
>        classifier and replicator as well?

No. There is no queue.


> (c) ...that the task of classifying packets according to their addr,
>        or 'replicating' a packet, does not have a time associated 
>        with it (instantaneous)?

No, It does not.


> (d) ...if (c) is 'true', there is no bound for the rate in which
>        a router could route/forward packets? (ignoring drops
>        caused by overflow of the outgoing queues into links?)

I don't know what you mean by 'forwarding packets' but they are certainly 
limited by the outgoing link speed. 

> 
> (f) ...in case of a host receiving packets, addressed to a local
>        agent, a transport layer sink. Is it possible that
>        packets (from multiple links) arrive at a host at 
>        a rate which is superior to the rate in which the
>        sink agent is receiving such packets? Were drops modeled,
>        in which queues that would be?
> 
The sinks can absorb the packets at any rate and hence the aggregate of 
all the incoming link rates. There are no buffers within the node and 
no drops are modeled within the nodes right now.



> (g) ...all buffer management is done in a packet-basis, not byte,
>        so that the packet size would not matter in regards to queue sizes
>        and drops?
> 
If you use a Drop Tail (FIFO) queue that is true. But if you use RED you have
an option of managing the buffer in Bytes. So the buffer management is specific
to the queue that you are using.
--------------
To summarize, Packets coming into a node enter at the node entry point. They
arrive at the incoming link speeds and are classified by the classifier to
local agents or appropriate outgoing link interfacess. There is no buffer/ 
buffer management between the node entry and classifier or between classifier 
and the agents.

The packets are assumed to have been absorbed instantaneously. The packets
to be forwarded to other nodes are done so also instantaneously to the link 
entry
point. (To my knowledge no delays are modeled here). 

The link object has a queue that stores the packets and forwards them
depending on the line rate and the propagation delay of the link that you
had specified. These are the only limitation for the forwarding of the packets.

Packets are managed by the queue objects (that are part of link objects). The 
drops are modeled here. (Overflow, early etc. depending on the queue 
management). Packets could also be dropped at the ttlchecker object (also part
of link object) if you are using the ttl field. The queue management can be done 
either in bytes or packets depending on the queue management that you are using.

good luck
kedar