A suggested guiding principle for 10GbE WAN design
All:
Now that Howard Frazier has come up with a good solution for separating the
10GbE WAN for carriers from the LAN issues, we still need a 10GbE WAN
solution for the rest of us.
The guiding design principle I would like to suggest is this: "Can the 18
year old kid who manages my LAN also manage my WAN? Can he use the same
test gear and management equipment for the LAN to manage the WAN?"
Let me you a real world example:
Many of our school boards and municipalities are deploying dark fiber
networks. To date they are driving the fiber with FastE or GigE switches
and/or transceivers. In many cases the links are only 1-5 km and so off the
shelf equipment equipment will work quite well. However, in every school
board or municipality there is usually one or two schools that are 10-50km
(sometimes 100-200km) away. These school boards and municipalities do not
want to have to buy special optical equipment, test gear and management
systems for these links. All they want is another blade or interface to
drive to be able to deliver data to those remote destinations. In some
cases they will have to install one or more intermediate regenerators
(usually just another switch). Again they will refuse to buy special test
gear and management equipment for just that one piece of equipment.
This model also extends to many of corporate campus participants as well.
In that case they usually have 2 separate drak fiber links for redundancy
and reliability. But again, they don't want technology that requires
special equipment or knowledge.
There has been a lot of discussion if using some form of SONET or SONET lite
for 10GbE WAN. That may be fine for the carrier market. But I doubt if it
will work for the scenario given above. More importantly there is a lot of
work in ITU and OIF and elsewhere to define that standard.
I think HSSG should concentrate on the technical space of extending the LAN
into the WAN and not the other way around. We need and want a technology
for the WAN that the 18 year old kid who runs my LAN can understand and
manage without the need of special training, test equipment and management
systems.
Fiber is cheap, cheap, cheap. Encoding efficiency is not a big deal for
this customer base. 8b/10b will be just fine thank you. So the closer we
can remain to the LAN standards the better.
When ( and if ) we have to connect to a carrier network to extend our 10GbE
extreme long distances will we require the special WAN 10GbE as discussed at
length on this list
Bill
Bill St. Arnaud
Senior Director Network Projects
CANARIE
bill.st.arnaud@xxxxxxxxxx
+1 613 785-0426
> -----Original Message-----
> From: Roy Bynum [mailto:rabynum@xxxxxxxxxxx]
> Sent: September 14, 1999 7:18 AM
> To: Bill.St.Arnaud@xxxxxxxxxx
> Cc: HSSG
> Subject: Re: Long distance links
>
>
> Bill,
>
> What sort of fiber fault management or maintenance without
> effecting traffic
> transport do these lasers provide?
>
> Thank you,
> Roy Bynum
> MCI WorldCom
>
> "Bill St. Arnaud" wrote:
>
> > Roy:
> >
> > Although there is no 802.3 WAN standard we do have GbE switches
> from vendors
> > that have 23 db lasers ( i.e. will drive over 100 km on G.652
> fiber). The
> > more common alternative is to use Gbe laser transceivers from a
> number of
> > suppliers. These also come with 20-23 db lasers
> >
> > No co-location space is required except at the node where the customer's
> > fiber interconnects to a DWDM channel, currently IP over SONET
> >
> > Shortly we are issuing an RFP for a "customer empowered
> network". Please
> > send me an e-mail for those who are interested in a draft copy
> >
> > Bill
> >
> > Bill St. Arnaud
> > Senior Director Network Projects
> > CANARIE
> > bill.st.arnaud@xxxxxxxxxx
> > +1 613 785-0426
> >
> > > -----Original Message-----
> > > From: Roy Bynum [mailto:rabynum@xxxxxxxxxxx]
> > > Sent: September 12, 1999 6:06 PM
> > > To: Bill.St.Arnaud@xxxxxxxxxx
> > > Cc: HSSG
> > > Subject: Re: Long distance links
> > >
> > >
> > > Bill,
> > >
> > > I believe that the "customer empowered networks" you are
> > > referring to, are those
> > > where the enterprise data customer is leasing the dark fiber from
> > > a utility or
> > > service provider. Because there isn't a WAN compatible PHY
> > > available for GbE,
> > > it requires a complete 802.3 data switch to regenerate the signal
> > > for long haul
> > > systems. It means that the enterprise data customer has to get an
> > > agreement from
> > > the utility or service provider to provide collocation space for
> > > the customer's
> > > data switch. It also means that the customer provides his own network
> > > management of the fiber facility. Over long distances, that
> could mean
> > > regularly dispatching someone to the various sites to support the
> > > collocated
> > > non-carrier data switches. While it is technically feasible to
> > > do this, over a
> > > period of time, this could get very expensive for enterprise data
> > > customers.
> > > Overall cost of ownership of this type of implementation does not
> > > make it very
> > > attractive. I do not think that there will be too many
> enterprise data
> > > customers that will be doing this for very long.
> > >
> > > Thank you,
> > > Roy Bynum
> > > MCI WorldCom
> > >
> > > "Bill St. Arnaud" wrote:
> > >
> > > > Roy, Rich:
> > > >
> > > > I think this reflects a fundamental misunderstanding between
> > > data people and
> > > > telco transport people.
> > > >
> > > > Yes, it is true that 10GbE data switch will not be satisfactory for
> > > > regeneration of traditional mission critical traffic on carrier
> > > networks.
> > > >
> > > > However, we believe there is a whole new set of traffic requirements
> > > > resulting from "customer empowered networks" where 10GbE data
> > > switches will
> > > > be more than adequate for regeneration.
> > > >
> > > > What we are starting to witness in the marketplace is the
> deployment of
> > > > these customer empowered networks. Some next gen carriers,
> > > for example do
> > > > not deploy SONET rings as a given network requirement. Instead
> > > rings and
> > > > the type of regeneration are a customer option.
> > > >
> > > > Bill
> > > >
> > > > Bill St. Arnaud
> > > > Senior Director Network Projects
> > > > CANARIE
> > > > bill.st.arnaud@xxxxxxxxxx
> > > > +1 613 785-0426
> > > >
> > > > > -----Original Message-----
> > > > > From: owner-stds-802-3-hssg@xxxxxxxxxxxxxxxxxx
> > > > > [mailto:owner-stds-802-3-hssg@xxxxxxxxxxxxxxxxxx]On Behalf Of
> > > Roy Bynum
> > > > > Sent: September 11, 1999 3:33 PM
> > > > > To: rtaborek@xxxxxxxxxxxxxxxx
> > > > > Cc: HSSG
> > > > > Subject: Re: Long distance links
> > > > >
> > > > >
> > > > >
> > > > > Rich,
> > > > >
> > > > > I am not writing that 10GbE end systems be managed by
> > > SONET/SDH network
> > > > > management standards. I wrote that trying to use 10GbE data
> > > > > switches as part of
> > > > > the transmission service systems would not meet the standards of
> > > > > the commercial
> > > > > services systems. The suggestion of putting a 10GbE switch in as
> > > > > a regenerator
> > > > > does not meet the security, operations support, or
> network management
> > > > > requirements that are in place for commercial transmission
> > > > > equipment. This is
> > > > > something that IP based data people normally do not have to deal
> > > > > with. It is
> > > > > easy to understand why they would not realize why a 10GbE data
> > > > > switch would not
> > > > > be used as a commercial service transmission regenerator. I
> > > was simply
> > > > > attempting to explain to them, with as few words as possible why
> > > > > this was so.
> > > > >
> > > > > Thank you,
> > > > > Roy Bynum
> > > > > MCI WorldCom
> > > > >
> > >
>