||: IP Over Sonet Article :||
All, The author of this article takes the liberty of regarding SONET as an optical layer protocol, but IMO, SONET is in actuality not purely optical, rather Opto-electronic, with convergence sub-layers mounted on top of the optical base. In any event, this is an interesting article on how IP router vendors are heading towards bypassing some of the middle steps.
In the author's mind, the dreaded ATM "cell tax" is depicted as onerous and burdensome. Is it any more or less burdensome than the slow-start algorithms of TCP/IP and the restarts necessary to counteract the failed "best-effort" attempts in the non-deterministic purely TCP/IP world? Any thoughts?
Some good coverage of Tag and MPLS thrown in as well.
Frank Coluccio
==============================
Data Nets Turn To Internet Protocol-Over-SONET
June 1, 1998
The pieces are falling rapidly in place to support an Internet
Protocol- dominated network infrastructure that is vastly
different from convergence- driven strategies that have
taken shape around existing data switching options,
including Asynchronous Transfer Mode and Ethernet
protocols.
In recent months, several major data networking
companies, including America Online Inc., GTE
Internetworking, Sprint Corp., Time Warner Cable and U
S West Media Group's MediaOne Express, have moved
to networks that transport packets directly over
Synchronous Optical Network (SONET) technology for
point-to-point transfer of volume data traffic, avoiding
the intermediate steps of Layer 2 switching such as
Asynchronous Transfer Mode (ATM) and Gigabit
Ethernet. And, because much of this traffic is mapped to
Internet Protocol (IP), these providers are positioning
themselves for future coupling of the Layer 3 IP traffic
control domain with the physical facility sub-layer
capabilities of all- optical networks, which would vastly
reduce the number of operating parts in any given
network.
The key to the new networking option lies in ever more
powerful and intelligent routers that are able to prioritize
IP data for classes of applications and to feed payloads
at speeds sufficient to fill OC-3, or
155-megabit-per-second, and OC-12, or 622-Mbps, pipes
now and much higher-speed SONET links later this year
and beyond. At the same time, with the advent of label
switching, where " tags" on packets eliminate the need
to read packet headers at every router, router
horsepower is being freed up to pass through higher
rates of data without sacrificing the ability to
differentiate traffic.
"The reality today is that Cisco [Systems Inc.] is running
OC-48 [2.5-gigabit- per-second] packet over SONET
connections [in demonstrations], and later this year we
will see a number of communications providers moving
to OC-48 backbones," says Graeme Fraser, vice president
of marketing at Cisco (www.cisco.com). "By the second
half of '99, we'll see a transition to OC-192 [10 Gbps]
based on packet-over-SONET."
Skip The ATM
Where ATM edge switches have been commonly used
to aggregate IP traffic from multiple routers at the
regional level for transfer to core switches via OC-3
pipes, the higher-speed capabilities of new routers
eliminate the need for the edge ATM interface, allowing
for end-to-end transmission via succeedingly higher
levels of multiplexing into SONET, Fraser says.
"Today this is a point-to-point technology, but there will
be an evolution to more complex architectures, starting
with rings and then moving to mesh topologies that use
Layer 3 routing [at all OC levels]," he adds.
This fits the course GTE Internetworking is on, where
IP-over-SONET is a solution for that Internet service
provider's (ISP's) IP data transport needs today and will
be used eventually for all categories of service.
"As we're building our new [national] fiber backbone,
initially we'll have separate circuits for voice, ATM and
frame relay," says Steven Blumenthal, vice president and
general manager at GTE Internetworking's
(www.bbn.com) global networking infrastructure. "But
over the next two to three years you'll see all that
converging [to IP]."
He adds: "We're relying on the router's ability to deliver
class of service and quality of service, to be able to mix
real-time traffic, like voice, video and multimedia, and to
give that an express lane through the network while
mixing that traffic with lower-priority data, all at the IP
layer."
GTE Internetworking, which supplies data access and
value-added services to end users and third-party
providers, has started down this path by using next-
generation routers from Cisco in conjunction with OC-3
and OC-12 SONET links, thereby avoiding the "cell tax"
of ATM, Blumenthal says. That tax, based on in- house
lab tests of ATM switches from a variety of vendors,
turns out to be approximately 17 percent of capacity for
GTE's mix of traffic, which consists of packets in the
range of 64 bytes in length and big Ethernet packets of
1,500 bytes.
"At the time we did the tests, we were leasing very
expensive circuits across the country and overseas, and
we couldn't afford to waste this bandwidth," he says.
The loss of efficiency in moving IP into ATM results
from the fact that fitting irregular length packets into
ATM cells, at uniform lengths of 53 bytes, often leaves
ATM cells only partially filled.
In addition, ATM devotes 5 bytes to the header in each
ATM cell that, combined with the IP header information,
results in bandwidth being wasted on redundant
information.
IP-SONET Mesh
AOL (www.aol.com), which is the largest data transport
customer for both GTE Internetworking and Sprint
(www.sprint.com), already has gone beyond the point-
to-point application of IP-over-SONET to the mesh
architecture approach within its data centers, says Victor
Parente, chief network architect at the service provider.
"We differ significantly from other ISPs in that we run
very large data center networks as well as transit
networks," he says. "When I started looking at IP-
over-SONET technology, what we were looking for was a
replacement for the current DS-3 (45-Mbps) technology
and ATM that we were using to aggregate intersite data
center traffic," Parente says. " There are a lot of problems
with [the ATM approach] in terms of the lack of maturity
of the software and also in the overlap or inconsistency
between routed IP architectures and switch
architectures."
The company decided to tear down the ATM complexes,
choosing instead to operate "clear channel" connections
between the centers using IP-over-SONET. Clear channel
links devote the entire bandwidth to the payload.
"All that complexity was replaced by a simple pair of
routers running back to back," Parente says. "Over time
we also saw the technology as useful in our peering
networks with our dominant Internet providers, and,
finally, we've also gotten to the point where we're using
IP-over-SONET really as a data center technology,
replacing LAN [local area network] technology. "
AOL is wrestling with issues that other providers of
mass-market services, such as high-speed cable and
Asymmetric Digital Subscriber Line, must deal with as
they ramp up to millions of customers.
"How do I bring in many gigabits of Internet capacity
into a data center? " Parente asks. "How do I design a
new network that's going to allow AOL to peer with
multiple ISPs?"
AOL must handle in excess of 10 Gbps per data center
and sustain high port densities connecting up to 64
routers, Parente says. "We needed redundancy and
reliability and what I call LAN/WAN [wide area network]
transparency, which I think is one of the really critical
values of IP-over-SONET," he says.
"This means that when I'm moving bits from one
computer room or data center to another, I want to be
able to take that data transfer on a continuous path
across the road or across the country without changing
data rates and protocols," he says. "I can use the same
capacity without having to deal with bottlenecks at some
point in my network."
Parente says he also looked at Gigabit Ethernet but
decided that "doesn't scale very well." A gigabit may
sound like a lot, he adds, "but my job is to plan further
into the future, and IP-over-SONET scales a little faster."
More Work Needed
As promising as the emerging IP-over-SONET strategy
is, there is much to be done to move this from a sort of
patched-together solution to one that specifically targets
the challenges of IP networking. One of the first tasks
concerns completion of the standard known as
Multi-Protocol Label Switching (MPLS), which is based
largely on the proprietary tag-switching system
developed by Cisco with still-to-be-resolved adjustments
as negotiated within the Internet Engineering Task
Force.
As things stand, use of the processing power in routers
to maximize data throughput exacts a certain amount of
compromise with regard to provisioning multiple classes
and qualities of service, says GTE's Blumenthal. His unit
is limiting itself to supporting three classes of service --
traditional " best effort" IP, a real-time class supporting
video and voice services and something in between to
support transactional services, where there is some
tolerance for response delay and packet "jitter," which is
a delay caused by packet dispersion in the network.
"The issue of why we don't do more classes of service
gets back to what routers can really do," Blumenthal
says. "When you turn on class of service in routers,
they start to spend cycles doing that instead of
forwarding packets. We really expect the router to fill
pipes at OC-48, so there's a little bit of a trade-off as to
what the router can really do."
GTE has been working "pretty closely with Cisco to
understand MPLS technology," Blumenthal says. "We
have some concerns with the current tag system, and
we're working very actively within the MPLS process
with the goal of deploying it in a more standard form."
The emergence of MPLS is a boon to the ATM world as
well, since it allows ATM switches to handle IP traffic
without having to apply redundant ATM header
protocols, says Edward Kennedy, vice president of
marketing at Alcatel Data Networks Inc.
(www.adn.alcatel.com) Earlier this month, Alcatel became
one of the first ATM vendors to implement tag
switching, promising to migrate to MPLS once the
standard is completed.
"As carriers begin to take data onto their backbones,
MPLS is an efficient, scalable way to support quality of
service," Kennedy says, adding that jumping the gun on
the standard makes sense, given Alcatel's long
relationship with Cisco and the fact that most of the
standard is based on tag switching. "Mapping Layer 3
[IP] onto ATM allows CLECs [competitive local
exchange carriers] and ILECs [incumbent local exchange
carriers] to support end-to-end networking at high
speeds while maintaining centralized management
control over the traffic."
In effect, MPLS allows big carriers with core ATM
switches to come as close as possible to the pure, Layer
3 connectionless network model without reconstituting
their entire architectures, Kennedy says.
"Carriers want to retain ATM because they risk having
to overprovision [bandwidth in] their network to
accommodate a purely connectionless environment, plus
there are issues of troubleshooting and billing control to
be resolved in the IP domain," he adds.
But, as these issues are worked out, and there is a crop
of software vendors in the IP telephony domain offering
many such solutions, the carriers may find the dynamic
bandwidth provisioning advantage of ATM becomes
outweighed by the "cell tax." This is because the
IP-over-SONET model is the beginning of a trend toward
direct connection of routers to the optical layer in
networks of the future, where Wave Division
Multiplexing (WDM) promises to free up a nearly infinite
amount of bandwidth.
"I believe that [IP-over-SONET] is one of the core
technologies that make up the optical networking
infrastructure of the future," says Cisco's Fraser.
"This is bringing routers and switches much, much
closer to the physical level, lining up directly with fiber
and feeding into the DWDM [Dense WDM] system." |
