[Netconf] NETCONF and interactive CLI's

Mahesh Jethanandani <mjethanandani@gmail.com> Mon, 08 September 2014 21:47 UTC

MIME-Version: 1.0
Date: Mon, 08 Sep 2014 14:47:25 -0700
Message-ID: <CAAchPMuqarL9LicAhWisZXSxk9=GYMsuxPVZcmMJ+dRjawzM6Q@mail.gmail.com>
From: Mahesh Jethanandani <mjethanandani@gmail.com>
To: Netconf <netconf@ietf.org>
Content-Type: multipart/alternative; boundary="001a1134b004b49dce050294c33d"
Archived-At: http://mailarchive.ietf.org/arch/msg/netconf/b0-ltZMzEB-dVE3k8eqR4b_Bwws
Subject: [Netconf] NETCONF and interactive CLI's
Precedence: list

I was asked to fashion a problem statement around how NETCONF deals with
what I call "interactive CLI" models.

That model includes commands that are today issued on a CLI and involve a
command being issued, and a continuous response coming back till such time
it is interrupted either by the CLI or because there is a terminating
condition. In the period that the command is running, the CLI may be
blocked from issuing any other commands.

Commands issued like these are fairly common debug commands issued as part
of testing the configuration on the box and are unlike getting statistics
which are not interactive in nature.

A very common example of this is the ping command. Other examples include
traceroute command. A successful ping would have a response like this.

 > ping google.com
PING google.com (74.125.239.98) 56(84) bytes of data.
64 bytes from nuq05s01-in-f2.1e100.net (74.125.239.98): icmp_req=1 ttl=55
time=4.38 ms
64 bytes from nuq05s01-in-f2.1e100.net (74.125.239.98): icmp_req=2 ttl=55
time=4.26 ms
64 bytes from nuq05s01-in-f2.1e100.net (74.125.239.98): icmp_req=3 ttl=55
time=4.38 ms

--- google.com ping statistics ---
3 packets transmitted, 3 received, 0% packet loss, time 2002ms
rtt min/avg/max/mdev = 4.260/4.343/4.389/0.096 ms

Here also the response is split in two parts, one that is interactive
response which spits out one line at a time as the ping reply comes back
and the second which for the lack of a better term a "terminating response"
- a cumulative response of the complete interaction, prompted by the
issuance of a terminating condition i.e. CTRL-C or TTL expiry.

In the case ping fails, there is no interactive response, but there is a
terminating response like in this case.

 > ping pyramid.com
PING pyramid.com (109.228.17.0) 56(84) bytes of data.


--- pyramid.com ping statistics ---
7 packets transmitted, 0 received, 100% packet loss, time 5999ms

Without getting into specific implementation details, these commands today
would follow a pattern of

<rpc>
...
<rpc/>

<rpc-reply>
....
<rpc-reply/>

<rpc-reply>
....
<rpc-reply/>

<rpc-reply>
....
<rpc-reply/>

<rpc>
CTRL-C for the above NETCONF session
<rpc/>

<rpc-reply>
....
<rpc-reply/>

How does NETCONF deal with such a command? I would not confuse this with
another thread that was talking about "bulk response" because this issue is
more than how to deal with fragments of large responses. It has to do with
the interactive nature that the current NETCONF protocol does not seem to
support. If it does not, what are semantics needed to support these kind of
commands.

-- 
Mahesh Jethanandani
mjethanandani@gmail.com

[Netconf] NETCONF and interactive CLI's Mahesh Jethanandani
Re: [Netconf] NETCONF and interactive CLI's Juergen Schoenwaelder
Re: [Netconf] NETCONF and interactive CLI's Andy Bierman
Re: [Netconf] NETCONF and interactive CLI's Kent Watsen
Re: [Netconf] NETCONF and interactive CLI's Andy Bierman
Re: [Netconf] NETCONF and interactive CLI's Juergen Schoenwaelder
Re: [Netconf] NETCONF and interactive CLI's Andy Bierman
Re: [Netconf] NETCONF and interactive CLI's Juergen Schoenwaelder
Re: [Netconf] NETCONF and interactive CLI's Martin Bjorklund
Re: [Netconf] NETCONF and interactive CLI's Martin Bjorklund
Re: [Netconf] NETCONF and interactive CLI's Bert Wijnen (IETF)
Re: [Netconf] NETCONF and interactive CLI's Mikael Abrahamsson
Re: [Netconf] NETCONF and interactive CLI's David Bannister
Re: [Netconf] NETCONF and interactive CLI's ietfdbh
Re: [Netconf] NETCONF and interactive CLI's Andy Bierman
Re: [Netconf] NETCONF and interactive CLI's Balazs Lengyel
Re: [Netconf] NETCONF and interactive CLI's Juergen Schoenwaelder
Re: [Netconf] NETCONF and interactive CLI's Romascanu, Dan (Dan)