Ali,
your points on the distance needs of
LANs and larger data centers resonate well with me. I have viewed
the 100m objective as a target that could be cost-effectively exceeded,
but also one that could get sufficient support to be accepted as an objective
by the HSSG and 802.3. Now we have seen at least one proposal for
specs that can exceed that objective by at least a factor of two on OM3
MMF. This is a positive indication that the final spec will exceed
the objective by wide margin. In addition to the Flatman data you
mention, both Steve Swanson and I have presented separate contributions
supporting longer distance needs in data centers. And, as we have
heard during the discussions on 4km vs 10km SM reach objectives, connection
insertion losses also need to be addressed beyond the 1.5 dB allocation
of previous efforts due to both the increased use of structured cabling
in data centers, which introduces central cross connection facilities,
and the deployment of preterminated cabling that is favored for new installations
in larger data centers, which doubles the number of connections per channel.
Under these considerations, specifications
that support at least 200m on OM3 with at least 3dB of connection loss
allocation would be optimal for data centers in my opinion.
It is an established trend that high
speed technologies are often first deployed in data centers then in LANs.
A 200m capability, while probably sufficient for DCs will not be
optimal for LANs due to the greater geographical coverage of whole buildings
compared to that of most DCs. To aid eventual LAN deployment, the
specifications should permit the support of the MMF solutions to at least
300m. One means of enabling this distance lies in the specification
of MMF more advanced than OM3. John Dambrosia has recently announced
receipt of the liaison letter on this subject from ISO/IEC that Alan Flatman
and I noted in our separate contributions at the November 802.3 meeting.
This letter requests feedback from 802.3 regarding the utility
of the proposed specifications for a higher performance MMF. Consideration
for the future needs of in-building LANs would be a reason to encourage
its standardization.
Regards,
Paul Kolesar
CommScope Inc.
EnterpriseŽ Solutions
1300 East Lookout Drive
Richardson, TX 75082
Phone: 972.792.3155
Fax: 972.792.3111
eMail: pkolesar@commscope.com
Ali Ghiasi <aghiasi@BROADCOM.COM>
12/21/2007 12:54 PM
Please respond to
Ali Ghiasi <aghiasi@BROADCOM.COM>
To
STDS-802-3-HSSG@LISTSERV.IEEE.ORG
cc
Subject
[802.3BA] Reach Objective for MMF
Hi
On this reflector we have had many discussion on the SMF reach
objective. But I want to bring
up your attention to the current MMF reach objective which is only 100
m.
High density data centers switches will be based on the QSFP (XFP width)
or the CSFP (~26 mm),
I do not see a viable serial 40G or 4x25G WDM solution in these form
factor in near future.
The current proposed 4x25G WDM will be housed in a double wide Xenpak
(~81 mm).
Currently for reach larger than 100 m you must switch to 100 Gig 4x25G
WDM! Unless you are
designing your switch for reduce capacity double wide Xenpak will not
plug in to a QSFP or CSFP!
From pure density perspective you will be doing worse than dual stacked
SFP+.
According to Allan Flatman study page 7
http://www.ieee802.org/3/10GMMFSG/public/mar04/flatman_1_0304.pdf
300 m covers the building backbone and for larger data center more than
100 m may be needed.
In Atlanta I showed how even with sub 10G laser "300 m link on OM3
fiber
with linear receiver has 1 dBo less penalty than 100 m link OM3 with
limiting receiver"
http://www.ieee802.org/3/hssg/public/nov07/ghiasi_01_1107.pdf
I would like to explore the possibility of increasing the reach
objective on OM3 fiber from 100 m to
220 m or 300 m. I do understand the draw back of ribbon cable, but
here
we have situation where
the 4x25G solution just doesn't fit in to a QSFP/CSFP. Maybe after
a
season on biggest looser it might fit :-)
