RE: Please help to clarify some things!
Roy,
Perhaps the telephone industry has not implemented it because the telephone
industry was already in existence and using management before Ethernet came
along?
Seriously I don't think you are going to get anywhere on this mailer by
complaining about the "unreliable delivery" of management information on
this mailer. Such answers as exist have been well worked out for data
networking equipment are in widespread use. Their study is not part of the
development of Ethernet MACs. I would suggest that you learn some more about
the data networking world as it operates today, and about switched Ethernet
LAN deployment in particular, in parallel with teaching the rest of us about
the telecommunications world.
You are missing Rich's point below. SNMP can communicate management
parameters that have been collected. What needs to be done, in the framework
of the Ethernet/LAN/IP/SNMP, world is to identify what information needs to
be collected and how it is collected, and then to cast that information in
the form of SNMP MIB so it can be transported using that protocol.
Mick
-----Original Message-----
From: owner-stds-802-3-hssg@xxxxxxxxxxxxxxxxxx
[mailto:owner-stds-802-3-hssg@xxxxxxxxxxxxxxxxxx]On Behalf Of Roy Bynum
Sent: Sunday, September 05, 1999 5:14 PM
To: rtaborek@xxxxxxxxxxxxxxxx
Cc: HSSG
Subject: Re: Please help to clarify some things!
Rich,
If the methodology that is used for Ethernet is less expensive, why has the
telephony
industry not implemented it? Could it be that the inband nature of IP based
SNMP
network management on a particular link introduces more problems than it
solves. SNMP
uses UDP, not TCP so it is an unreliable information delivery system. Could
it be
that the unreliable delivery of alarm and fault information is considered
unreliable.
If you have done any research, you would know that IP based out of band
network
management for telephony systems is based on TCP, UDP. TL1 over IP uses
TCP.
Ethernet does not gather any information at the optical level. The lowest
common
denominator for Ethernet network management is the "frame" not the optical
path or
signal. As such, any SNMP management system that is dependent on Ethernet
as it is
currently implemented does not have visibility to the optical level, not
even
indirectly.
Thank you,
Roy Bynum
MCI WorldCom
Rich Taborek wrote:
> Roy Bynum wrote:
>
> > Mick,
> >
> > The splice points are just physical joinings of the fiber. The problem
is that
> > they can be inconsistent to the point of effecting the functionality of
the data
> > transmission. They can also be effected by people messing with them in
the
> > process of working on the fiber cable. As a cable degrades or is
damaged the
> > link level may fail to the point of not being able to use "ping", yet
the remote
> > status information can inform you of whether the problem is in both
directions
> > of the traffic, or just one, and which one. A ping can not tell you if
the link
> > failed in one direction only. As a transmission laser starts to fail,
the remote
> > system returning its receive status information can cause an allert/trap
against
> > the local system, even though the problem was seen on the remote system.
>
> Roy,
>
> Ethernet is at capable as SONET, at detecting the same conditions.
Management
> protocols above Ethernet are, in general, more capable of assisting
service
> personnel in maintaining links. I see no SONET advantage here. If
anything, the
> Ethernet advantage, once again, is lower cost.
>
> > At present, SNMP can only tell you if there are excessive error frames.
It does
> > not have visibility at the optical level.
>
> I believe that SNMP stands for something like Simple Network Management
Protocol.
> Ethernet has visibility at the optical level. Ethernet gathers information
at the
> optical level and SNMP presents the information for processing by a
management
> application. Therefore, SNMP indirectly has visibility at the optical
level.
>
> > Having a network management system be
> > able to properly report where the problem is will save a lot of time,
effort,
> > and expensive support people in the process of resolving the problem.
Part of
> > what I am recommending is adding to SNMP the visibility to the optical
level.
>
> Too late.
>
> > Thank you,
> > Roy Bynum
> > MCI WorldCom
> >
> > Mick Seaman wrote:
> >
> > > Roy,
> > >
> > > I'm sorry, I feel you are repeating "full duplex 10Gb 802.3 is in need
of
> > > protocol
> > > level operations support functionality" but not adding to my
understanding
> > > of why at all. I am probably not alone.
> > >
> > > Can we be more specific about those things that the functionality you
> > > propose will enable us to diagnose that can not be accomplished by
> > > collecting data in the systems attached to the Gigabit MAC and then
sharing
> > > that data or responding to queries using that MAC to transmit and
receive.
> > >
> > > Is it possible to see or manage the splice points you refer to using
an
> > > embedded protocol? I thought they were just physical splice points and
that
> > > none of the protocols you were discussing contained embedded OTDR. So
the
> > > only thing that would help me to manage these would be better insight
into
> > > BER or 'physical' level signal conditioning at the end stations. Why
can't I
> > > get at the information provided by this using SNMP, or check
connectivity
> > > using 'ping' etc. etc. What is there here that needs support in the
MAC? It
> > > may be that the world needs better management protocols but why is
that a
> > > subject for HSSG discussion?
> > >
> > > Mick
> > >
> > > From: owner-stds-802-3-hssg@xxxxxxxxxxxxxxxxxx
> > > [mailto:owner-stds-802-3-hssg@xxxxxxxxxxxxxxxxxx]On Behalf Of Roy
Bynum
> > > Sent: Sunday, August 29, 1999 10:24 PM
> > > To: Henry Ngai
> > > Cc: stds-802-3-hssg@xxxxxxxx
> > > Subject: Re: Please help to clarify some things!
> > >
> > > Henry,
> > >
> > > In your document, your first and second drawings are somewhat correct.
I
> > > doubt
> > > that 10GbE would ever be implemented as you are showing it in your
third
> > > drawing. In your first drawing, you need to add a fiber plant that
has
> > > things
> > > like splice points every 5 km at most, access to the fiber cable by
people
> > > that
> > > have nothing to do with your data, and other issues. The traditional
802.3
> > > LAN
> > > environment was outgrown when full duplex 802.3 over optical transport
was
> > > standardized. It is even possible to take full duplex 100BaseFX over
WAN
> > > distances with optical converters that are sold by several different
> > > vendors.
> > > As seen by individuals that are attempting to create manageable
enterprise
> > > level data networks with GbE, full duplex 10Gb 802.3 is in need of
protocol
> > > level operations support functionality in order to grow to its true
> > > potential.
>
> --
>
> Best Regards,
> Rich
>
> -------------------------------------------------------------
> Richard Taborek Sr. Tel: 650 210 8800 x101 or 408 370 9233
> Principal Architect Fax: 650 940 1898 or 408 374 3645
> Transcendata, Inc. Email: rtaborek@xxxxxxxxxxxxxxxx
> 1029 Corporation Way http://www.transcendata.com
> Palo Alto, CA 94303-4305 Alt email: rtaborek@xxxxxxxxxxxxx