Re: [802.3ae_Serial] Support for 10GFC
Pat, Vipul,
On the protocol side of the issues presented here, 10 Gigabit Fibre
Channel uses the special characters K28.2 and K28.4 through the XGMII,
8B/10B PCS and 64B/66B PCS. 10GFC does not use K28.6.
We're still working the link budget issue but the direction of comment
resolution thus far has been to reduce 10GFC PMD supported distances
where link budget unallocated margin is not available. Since the
difference is 2% between 10GFC and 10GE I believe that this solution
will be acceptable to FC customers. One has to consider the 10G
convergence where SONET OC-192, 10GE, 10GFC and FEC versions yield
operating rates of 9.952 Gb/s though 12.5 Gb/s for serial and 2.488 Gb/s
through 3.9 Gb/s for 4 lane (also includes InfiniBand). The 2% 10GFC
variation from 10GE literally gets lost in the noise.
Happy Holidays,
Rich
--
"THALER,PAT (A-Roseville,ex1)" wrote:
>
> Vipul and Tom,
>
> The 10GFC draft says in an editor's note at the start of the specs for the
> PMDs they have borrowed from Ethernet:
> The 10GFC project intends to compensate for the higher speed by making
> compensating adjustments in specifications relating to cable length and
> optical link budgets.
>
> They are aware of the issue and have decided to deal with it by adjusting
> the link rather than the transceiver.
>
> If you look at the specs for XGXS and PCS in 802.3ae, they actually do
> include requirements to properly handle the reserved codes of K28.2 and
> K28.6 (see Tables 48-2, 48-3, and 49-1). There was a brief period when one
> of the tables in 48 indicated otherwise, but I submitted a comment to
> correct that which was accepted. The physical layer components defined in
> 802.3ae will handle the FC encodings. However, I think your statement is
> partially incorrect as K28.6 is listed as reserved in the 10GFC draft. I
> believe that K28.2 is the only code additional that 10GFC uses.
>
> I am in the process of reviewing the specs in 10GFC to see if there are any
> deviations that would interfere with using the same transceiver or XGXS for
> both standards. I encourage others to do the same.
>
> Regards,
> Pat
>
> -----Original Message-----
> From: Vipul Bhatt [mailto:vipul.bhatt@xxxxxxxxxxx]
> Sent: Tuesday, December 18, 2001 10:06 AM
> To: 802. 3ae Serial PMD (E-mail)
> Subject: RE: [802.3ae_Serial] Support for 10GFC
>
> Tom,
>
> Good idea. From the standpoint of return on development efforts,
> 10GFC is a low hanging fruit. The increase in power budget is
> relatively small.
>
> I hope the designers of XGXS and PCS will do a similar exercise. My
> (unreliable) recollection is that apart from the increased 2% speed,
> the only thing a designer has to make sure is that K28.2 and K28.6
> are supported. Support for these two characters is required in
> 10GFC, but not in 802.3ae.
>
> Regards,
> Vipul
>
> =================
>
> -----Original Message-----
> From: owner-stds-802-3-hssg-serialpmd@xxxxxxxxxxxxxxxxxx
> [mailto:owner-stds-802-3-hssg-serialpmd@xxxxxxxxxxxxxxxxxx]On Behalf
> Of Lindsay, Tom
> Sent: Monday, December 17, 2001 4:01 PM
> To: 802. 3ae Serial PMD (E-mail)
> Subject: [802.3ae_Serial] Support for 10GFC
>
> Folks - 10G Fibre channel intends to use 10G Ethernet PMDs but for a
> rate that is 2% faster. For serial, this would mean 10.51875 Gbd;
> for LX4, this would mean 3.1875 Gbd per lane.
>
> As a modeling approximation, I modified cell C4 in
> 10GEPBud3_1_16a.xls to the FC rates. Margins stay positive for
> all -S variants, but go negative for:
> -0.05 dB for 1310 serial
> -0.21 dB for LX4
>
> Possibly any of these are within spreadsheet margin for error, but
> is there willingness to increase the power budget to drive these
> margins back positive?
>
> Tom
> Stratos NW
> 425/672-8035 x105
---------------------------------------------------------
Richard Taborek Sr. Intel Corporation
XAUI Sherpa Intel Communications Group
3101 Jay Street, Suite 110 Optical Strategic Marketing
Santa Clara, CA 95054 Santa Clara Design Center
408-496-3423 JAY1-101
Cell: 408-832-3957 mailto:rich.taborek@xxxxxxxxx
Fax: 408-486-9783 http://www.intel.com