>Cool, thanks for the confirmation and the education.
>
>Funny thing, I'm using your Designing Routing and Switching book as one of
>my references.
>
>Thanks again,
>Sean C.
>CCNP, CCDP, MCSE
One who references such works can do no wrong. :-)
Seriously, I had been about to follow up anyway. You make the valid
point that there can be conflicting opinions in books and review
guides, which often come down to opinion.
While the Internet is, to some extent, a matter of opinion, there are
some reasonably definitive things to look at. Admittedly, they are
written for protocol developers, and may not be the most readable
things in the world. But above all, look at RFC 1812, "Requirements
for IPv4 Routers," which takes a different perspective than the
individual protocols -- instead, it deals with what this thing called
a router is supposed to do.
The editor of 1812 is Fred Baker, who just stepped down as the very
long-term chair of the IETF. I was at the "changing of the guard" at
the Minneapolis IETF last month, and I have to say there was a lot of
emotion in the room as Fred turned it over to Harald Alvestad. Fred
done good. Fred is also a Cisco employee, so 1812 should generally
be consistent with Cisco usage.
The Internet doesn't have a formal architectural specification in the
same sense as does OSI, but a great deal of underlying work gets
pulled together in RFC 1958, "Architectural Principles of the
Internet." Brian Carpenter edited it for the IAB.
At the level of the IAB, it will be freely admitted that there are
lots of architectural violations in the real world, many of which
serve useful purposes. You can follow the ongoing discussions in the
IETF, in such groups as midbox (the broad implications of NAT,
proxies, tunnels, etc., as opposed to the specific protocols of each
of these) and the sub-IP temporary area (dealing with the
relationship of IP to specific transmission systems and things like
MPLS).
>
>
>""Howard C. Berkowitz"" wrote in message
>[EMAIL PROTECTED]">news:[EMAIL PROTECTED]...
>> >Hi All,
>> >
>> >HostA----------Router1-----------Router2-----------HostB
>> > Ether Serial Ether
>> >
>> >If a packet is sent from HostA to HostB and the packet is lost between
R1
>> >and R2 - who resends?
>> >
>> >I've seen this question before and thought the generally agreed upon
>answer
>> >was that Host A would resend with the reason being "The data-link layer
>> >protocols in use today on serial lines, including PPP, Frame Relay, and
>> >Cisco HDLC, detect errors but are not responsible for error correction
or
>> >retransmission."
>>
>> In general, I'd say Host A. The difficulty, however, is that there
>> is no absolute rule for what hosts do and what routers do. It's a
>> question of the specific protocols in use.
>>
>> If, for example, the Ethernet between R1 and R2 were running X.25
>> (not a common configuration, but specified in ISO 8880), R1 would
>> retransmit.
>>
>> I can come up with even weirder configurations involving multiple
>> levels of tunneling, but, if it is an IP environment, hosts are
>> architecturally responsible for the reliability of communications. I
>> used words very carefully there -- I'm not saying "reliable" but
>> simply the appropriate level of reliability. If, for example, the
>> hosts are running a UDP-based application such as streaming video, no
>> one will retransmit.
>>
>> Again, the IP architecture assumes the default behavior of
>> retransmission, if done at all, is done in hosts. Routers retransmit
>> only when they are using data link protocols that specifically
>> retransmit (e.g., SDLC and LAP-B), or when they are acting as a host
>> (e.g., as a protocol converter or endpoint of a reliable tunnel).
>>
>> >
>> >The reason I post this is I have found a contradictoring document and
>wanted
>> >some 2nd and 3rd opinions. The document in questions is the cramsession
>> >from BrainBuzz.com at the link:
>> >
>> >http://cramsession.brainbuzz.com/cramsession/cisco/ccie_written/
> > >
>> >The statement in question is at the top of Page 13. It states the the
>> >packet would be resent (actually rebroadcast) from Router B.
>> >
>> >I'm not using this guide to learn from as much to reinforce what I have
>> >learned and this statement is bringing up an old question.
>> >
>> >Thanks to all who answer,
> > >Sean
Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=671&t=662
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
