Re: [802.3BA] PHY nomenclature letters
David
All this alphabet soup is going to confuse customers.
I might be time to come up with a new paradigm, syntax that could carry
forward for 40, 100, 400, 1000G.
No one was ever confused by OC-768 VSR. Perhaps dot3 should innovate by
borrowing from the SONET nomenclature that separated the signifier for
the interface type/coding from the signifier for the optics type.
I am not sure E for 40 km will have much traction in this segment since
once you leave the data center, the links should be engineered just like
10GBASE-ER links (and the various proprietary 10G wavelengths that are
on the market). Even 10GBASE-LR does not run 10Km on some of the
installed SM fiber.
My 2 cents
Bruce
Bruce Tolley
VP, Corporate Marketing
Solarflare Communications
mobile: 650 862 1074
ip phone: 949 581 6830 x 2014
-----Original Message-----
From: David Law [mailto:David_Law@xxxxxxxx]
Sent: Sunday, July 12, 2009 10:11 AM
To: STDS-802-3-HSSG@xxxxxxxxxxxxxxxxx
Subject: [802.3BA] PHY nomenclature letters
All,
There has been a lot of discussion over comment #255
[http://www.ieee802.org/3/ba/public/jul09/P8023ba-D21-Comments_Received_
byCls.pdf#Page=4]
that Brad submitted against IEEE P802.3ba D2.1 in relation to consistent
use of nomenclature letter. Due to this I tried to record where I think
we
are in respect to PHY naming in IEEE 802.3, I have attached this
information at the end of this email and summarized in the attached pdf.
Based on this I'd say that I do not believe we have ever been
particularly
consistent with PHY nomenclature from project to project - and instead
we
have 'evolved' it where we agree there is a need to do so - which I
think
is reasonable. For example we moved from just indicating media ('F' for
fibre) in 100 Mb/s to wavelength ('S' and 'L' for short and long
wavelength) in 1 Gb/s since we had more than one fibre PHY. Further we
have allowed a nomenclature letter to have different meaning dependant
on
position ('S' and 'L' mean short and long reach as a second letter where
encoding is not indicated). What however I do not believe we have ever
done is create what I will characterise as a conflicting definition -
where the same letter in the same position means something different.
Based on the above I believe that comment #255 points out that IEEE
P802.3ba will result in a conflicting definition in IEEE 802.3. I
believe
that this is acknowledged in IEEE P802.3ba D2.1 subclause 80.1.4
'Nomenclature' since it states the existing IEEE Std 802.3-2008 meaning,
then redefines the meaning for 40Gb/s and 100Gb/s, of the first letter
'S', 'L' and 'E' (from Short, Long and Extra long Wavelength to Short,
Long and Extra long Reach).
In respect to the suggested remedy, to change 100GBASE-ER4 to
100GBASE-HR4
using 'H' to indicate high-power (or higher sensitivity) long
wavelength,
while not in conflict with what has been done in the past, would I
believe
evolve what the first letter has been used for up to now (Media or
Wavelength) and I'm not personally convinced of the need to do that in
this case.
Instead I suggest an alternative approach to the 100GBASE-ER4 PHY name.
With the first letter after the dash based on either the Media or
Wavelength the 'E' should be changed to an 'L' as it is a 1310 nm (long
wavelength) based solution. The second letter either indicates encoding
or, if encoding is not indicated, reach. Based on this I believe it
should
remain 'R'. This would result in no differentiation between this 40 km
PHY
and the 10 km single mode PHY so additional differentiation is required.
In the IEEE 802.3ah project this was dealt with by adding a number (e.g.
1000BASE-LX10), but we can't do this here as at speeds of 10Gb/s and
above
the number is used to indicate the number of lanes.
I therefore propose we add a third letter, possibly 'E' for extended
reach. I propose this to partially address the concern the Task Force
has
that the market associates E with 40km which John D'Ambrosia pointed out
to me. The 100GBASE-ER4 PHY would therefore be renamed 100GBASE-LRE4
PHY.
Based on that I suggest that the comment can be entirely satisfied by
two
actions:
[1] Deleting the contents of subclause 80.1.4 'Nomenclature' that states
that the meanings of S, L and E are being changed and making changes as
required to align to this.
[2] Renaming the 100GBASE-ER4 PHY to be the 100GBASE-LRE4 PHY.
One final thought. We may not have done a fantastic job of formally
recording what these letters and numbers actually meaning. As an example
while we do state that in subclause 1.4.35 '10GBASE-L' that it is a '..
PMD specifications for 10 Gb/s serial transmission using long
wavelength.'
we don't actually ever say that 'L' as a first letter means Long
Wavelength and as a second letter means Long Reach. I think that has
contributed to a situation where there are some, including participants
in
IEEE 802.3, that are of the opinion, for example, that Long reach, Long
wavelength and the first letter 'L' are synonymous.
Moving forward there may be a benefit to recording the entire
nomenclature
but I believe that is out of scope of the IEEE P802.3ba project. I
believe
however that Wael plans to add an agenda item to the Maintenance Task
Force to discuss this.
Best regards,
David
The following is a more detail description of what I show in the slides.
With the exception of the FOIRL MAU (which is now deprecated) I'll break
the naming down into three portions as follows:
<n><TYPE><ADD>
Where:
n is MAU/PHY data rate.
TYPE is MAU/PHY modulation type.
ADD is MAU/PHY additional distinction (which is broken down further
below)
n - data rate
=============
This defines the data rate that the MAU/PHY can operate and consists of
an
integer followed conditionally by a multiplier. Where only one number is
used the MAU/PHY operates at the same data rate both downstream and
upstream, when two numbers are used separated by a forward slash the
first
number defines the downstream (from OLT to ONU) data rate, the second
number defines the upstream (from ONU to OLT) data rate. Where the
multiplier is not used the number(s) define(s) the data rate in Mb/s,
where the optional multiplier is 'G' the number(s) define(s) the data
rate
in Gb/s. Data rates in use so far are:
2 2Mb/s
10 10Mb/s
100 100Mb/s
1000 1000Mb/s
10G 10Gb/s
10/1G 10Gb/s downstream, 1Gb/s upstream
Notes - (a) This is the data bit rate, not the Baud (encoded data) rate.
(b) In the case of IEEE 802.3ah Ethernet in the First Mile copper PHYs
this is the nominal maximum data rate since these PHYs can operate at
multiple bit rates (see 2BASE-T and 10PASS-TS, Clause 61, 62 and 63).
(c)
I've included the designations added by IEEE P802.3av 10Gb/s EPON as the
last recirculation just closed with 100% approval and no comments.
TYPE - modulation type
======================
The modulation type indicates how encoded data is transmitted on the
medium. Modulation types in use so far are:
BASE Baseband
BROAD Broadband
PASS Passband
Notes - Before Hugh points it out I acknowledge that fibre optics
MAU/PHYs
are actually modulating a carrier of a few hundred THz - however lets
not
go there please :-)
ADD - Additional distinction
============================
There are two basic formats that the additional distinction can take.
The
first, which is a legacy format, consists simply of a number after the
modulation type. This format can be summarized as (m). The second
consists
of a dash followed by one to three letters, conditionally a number,
conditionally a dash and a letter, then conditionally a number. This
format can be summarized as (-LLLm-Eo). These are both described in more
detail below:
m format additional distinction
-------------------------------
Where the additional distinction consists of digits directly after the
modulation type the digits define the maximum segment length, rounded up
to the nearest 100 meters, divided by 100 meters. An example is 10BASE2
which has a maximum segment length of 185 meters. This is a legacy
format
that was only used for 10Mb/s Coax MAUs.
-LLLm-Eo format additional distinction
---------------------------------------
Where the additional distinction starts with a dash there will follow a
minimum of one, and up to three, letters. The meaning of the letters are
position dependant and those in use today are defined as follows:
First letter (media or wavelength)
B Bidirectional optics
C Twin axial copper
E Extra long wavelength (1550nm)
F Fiber
K Backbone
L Long wavelength (1310nm)
P Passive optics
S Short wavelength (850nm)
T Twisted pair
Second letter (reach or PCS encoding)
B Backbone
L Link (10BASE-FL), Long reach (2BASE-TL)
P Passive optics
R Scrambled coding (64B66B)
S Short reach
W WAN coding (SONET/SDH)
X External sourced coding (block coding included by reference, 4B5B,
8B10B)
Third letter
M Multimode
X External sourced coding (block coding included by reference, 4B5B,
8B10B)
Where two encoding are included the first letter represents the
downstream
encoding, and the second letter represents the upstream encoding. An
example is 10/1GBASE-PRX-D1 where 64B66B scrambled encoding is used
downstream while 8B10B block encoding is used upstream.
For optical PHYs where the data rate is 1000 Mb/s or less, if number (m)
is present, it represents the maximum segment length in km. For Twisted
pair PHYs where the data rate is 100 Mb/s, if number (m) is present, it
represents the number of pairs used. Where the data rate is 10Gb/s or
more
the number (m) represents the number lanes, the absence of a number
indicating single lane operation. Examples are 100BASE-LX10 which has a
maximum segment length of 10 km, 100BASE-T4 which operates over 4 pairs
and 10GBASE-LX4 which operates over four lanes.
For asymmetrical PHY types the dash and letter (-E) indicates if the PHY
is a network-end (OLT, Central office) or subscriber-end (ONU) PHY. The
values currently used to indicate the end are:
D Downstream (OLT)
U Upstream (ONU)
O Central Office
R Subscriber
For Passive optical PHY types that support multiple optical power
budgets
the number (o) attached to the end indication indicates which of the
power
budget the PHY supports. The values currently used to indicate the power
budget are:
1 10 km and a split ratio of at least 1:16
2 10 km and the split ratio of at least 1:32
3 20 km and a split ratio of at least 1:32
Note - The PCS and PMD family names based on use of either the first or
second letter. Examples are 10GBASE-L for 10Gb/s long wavelength PMD
family and 10GBASE-R for 10Gb/s scrambled encoding PCS family.