National LambdaRail Network
The National Lambda Rail, as described in an email from Ron Johnson.
Here is a good description of the National Lambda Rail Network.
In NM the Lambda Rail Network is the fiber backbone connecting a number
of R&D labs (Los Alamos, Sandia) and universities, as well as State
fiber (and microwave) backbone systems; with intentions for greater
network outreach connections.
rl
---------- Forwarded message ----------
Date: Wed, 13 Sep 2006 10:59:33 -0700 (PDT)
From: Ron Johnson <[email protected]>
I am a co-founder, and a board member of National LambdaRail
(NLR).
Organizationally NLR is a not-for-profit consortium with 501c3
status whose 'members' are mostly the major research and education
(R&E) 'regional networks' (like SoX in Georgia) and 'regional
optical networks' like CENIC in California, FLR in Florida, and LONI in
Louisiana (aka 'RONs' because they are based on 'owned' and self-lit
fiber optic infrastructures). These R&E 'regionals' and 'RONs'
typically connect most of the major research universities and labs
(sometimes including some corporate research facilities), plus often
entire statewide (or even multiple state's) 'k20' networks (often
including all of higher education as well as most/all of public k-12,
and sometimes museums and other cultural institutions) to each other,
to costeffective backbone level commodity internet services, and most
importantly to the the rest of the R&E community via major research
and education national backbones like NLR, Internet2/Abilene, ESnet,
UltraScience Net, and DREN etc., as well as to various domestic and
international R&E peering and exchange points.
NLR is a particularly interesting version of a national 'R&E'
backbone as it is the only one in the USA that is based upon NLR having
actually 'bought' (in reality this means NLR has acquired 20 year IRU's
for) a national fiber optic cable footprint (and here mostly on
Level(3) System's fiber footprint) of roughly 15,000 miles of fiber
pair that interconnects the national major telecommunications hubs
(which is where the regionals and RONs home) and which connects the
'regionals' and 'RONs' to each other with 'owned fiber'. NLR 'lit' the
fiber with standard 'dense wave division multiplexing' (dwdm) 'optical
kits' (in collaboration with Cisco) that can provide dozens of separate
10 gigabit 'wavelengths' (aka 'lambdas') that are (as they are in such
systems) entirely independent from one another. Each individual
wavelength can be used, as some are, for normal high performance layer
3 (and also layer 2) internet services. Specific waves are also be used
as part of network and non-network research activities to run new and
different protocols and/or to provide specialized, necessarily
dedicated, services between specialized scientific
instruments/apparatus and/or compute, mass store and/or viz.
resources.
Among the many good things about having an optical system like this
with multiple independent lambdas (thin of them as separate
wires/circuits) is that one can run experiments that entail 'breakable'
networks in one wave (or 'lambda') in such an optical system along side
other completely independent waves that must support bulletproof
'production' network services for hospitals and clinics, without the
two divergent networks and needs affecting each other in the least
bit.
Another nice facet of such networks is that they can be (and in
cases like NLR are) configured to use the kind of technology and
provide the kind of services that you want and need, rather than just
what telcos and isp's are willing to sell. Here for example various
waves segments (and entire waves around the country) are configured as
plain 10 gigabit *ethernets* (ie as LAN PHY) rather than SONET
services, making it much easier and cheaper to build and operate
certain kinds of networks needed to support researchers and/or specific
R&E network service needs.
Another good thing about this sort of owned and lit fiber
infrastructure is that it provides the R&E community with 'actual
cost' based access to waves and bandwidth, and that is wildly less
expensive than even heavily discounted educational pricing from
carriers.
NLR's roots and focus to date have primarily been oriented to serving
research. Here that
especially includes both 'network research' and also (and quite
differently) specialized networks that support specific sorts of
research and education and R&D. That said, NLR's capabilities do
have broader utility and possible application in providing inexpensive
and very high performance general layer3 (and/or layer 2) network
services, as well as dedicated waves/'circuits', for various critical
mass communities (like k20 networks) for both intra-community network
services, and also for very cost-effective access to a powerful
combination of peering and exchange capabilities (including direct
settlement-free exchange with .coms like Microsoft, and with other
exchanges) and very cost-effective access to top tier ISP services for
remaining external traffic.
The NLR pages pretty much sketch the big picture, although in
R&E network community terms. If anyone has a specific
interest in NLR please feel free to contact me directly outside the
list.
NLR is an interesting infrastructure and one which at minimum we
hope will help ensure that R&E does not fall complete victim to the
attempted re-monopolization of telecom and internet facilities and
services. And in my experience peering and exchange fabrics
are, work, and are often treated much differently.
And, here in some areas smaller commercial telcos, pud's, coops,
cableTV and wireless
'operators' sometimes actually work together and with the R&E and
other not-for-profit net communities (especially including many of the
'regionals' and RONs (which i will add
sometimes call themselves Gigapops) to achieve mutual advantage (vs the
rboc's and major ISP's) in cooperating on peering and exchange
connections, sites, and fabrics; and even on cooperative fiber etc
projects.
I should probably also note that NLR has been set up to, and has a
rule set that encourages work & teaming with the private sector.
Further NLR is largely 'aup-free' in the sense that it does not
restrict commercial use or commercial collaboration (although 501c3
preservation and related 'UBIT' sorts of issues do need to be
considered and handled and NLR would not want to move towards serving
as any sort of commercial style ISP or commercially oriented layer3
transit service).
Ron Johnson
University of Washington,
Pacific Northwest Gigapop, Pacific Wave