Net World: Why ATM/IP QoS is killing Frame Relay
Frame relay at the crossroads
Frame Relay Forum mulls shift in focus.
BY TIM GREENE
Network World, 05/31/99
At the tender age of 8, the
Frame Relay Forum is having
an identity crisis.
In August, the forum's technical committee will
decide whether it should continue to develop
and set standards for the versatile packet
technology or scale back and acknowledge that
the forum's work is substantially done.
Instead of fading away,
proponents say the
forum should focus on
developing new
guidelines, such as how
frame relay should
interoperate with
emerging IP
quality-of-service (QoS)
technologies, including
Multi-protocol Label
Switching (MPLS).
ATM already surpasses
frame relay at supporting
QoS and makes it
possible to converge
voice, data and video on
one network. Even forum
members wonder if it
makes sense to keep
expanding frame relay
features in the face of
more fully featured
alternatives.
Lori Dreher, president of
the Frame Relay Forum,
says: "We don't have a
true frame relay class of
service or QoS, although
some vendors do similar
things that are
proprietary. The forum
certainly hasn't dealt with
it from a technical
perspective. Whether we
should is one of the
questions."
"The kind of work we are
doing maybe doesn't
need to be done in a
forum environment and is
better off left for
vendor-specific
implementations," says
Doug O'Leary, the
Frame Relay Forum
technical committee
chairman.
Given that such doubts exist, O'Leary has
asked the general membership whether the
technical committee should scale back its work,
go into sleeper mode, as O'Leary calls it, or
forge ahead.
The answer is clearer to others. "The forum is
getting bogged down trying to find busy work to
do," says Liza Henderson, senior broadband
consultant with TeleChoice, a telecom research
firm in Boston. "Frame relay has matured to the
point where it is a technology that is well
understood and implemented."
Analysts say frame relay will continue its heady
growth for a while, but the analysts see a
plateau on the horizon.
"In another couple of years frame relay will
reach its peak, stabilize for a couple years, then
go into decline," says Steve Sazegari, principal
with Tele.Mac in Foster City, Calif.
As countries with poor public network
infrastructures modernize, frame relay may
never be a factor, he predicts. ATM or
developing optical technology such as packet
over SONET will leapfrog frame relay, Sazegari
says.
One sign that the forum is at a turning point is
the recent difficulty the technical committee had
reaching consensus on multilink frame relay -
the ability to logically bond two separate frame
relay access circuits.
"A number of members argued [multilink frame
relay schemes] should be vendor-specific. An
equal number argued they wanted an
implementation agreement to standardize it,"
O'Leary says. The committee finally decided to
bring the proposal up for a vote this fall by the
full forum.
The struggle may indicate a change is needed.
"It's a waste of time to spend months working
on an implementation agreement that never
gets implemented," he says.
The forum might do better determining how
frame relay fits in with other technologies, says
Tom Nolle, president of CIMI Corp., a
technology assessment firm in Voorhees, N.J.
Network executives may want to use frame
relay switches as part of a network using MPLS
to speed traffic.
"That could rightfully be addressed by the
Frame Relay Forum. The Internet Engineering
Task Force could address it, but the answer
wouldn't be optimized for frame relay," he says.
Also, the question of how to bind together IP
and frame relay networks will inevitably become
an issue, so the forum could spell out how to do
it ahead of time using established standards,
Nolle says. The forum could specialize in writing
such application notes, he says.
But Nolle says it would be a mistake to disband
because that would signal the technology is in
decline, which it isn't.
"It's the most successful new service we've ever
had. There is more revenue from frame relay
than from the Internet," Nolle says. International
Data Corp. projects the sale of frame relay
ports will grow an average 15% per year
between 1997 and 2003.
And frame relay is key to supporting new IP
services today. AT&T's IP-Enabled Frame
Relay service mixes IP routing with frame relay
virtual circuits using MPLS. That lets all sites on
a network connect to all other sites without
needing to buy a full mesh of virtual circuits.
Those types of IP virtual private network
(VPN)services will ultimately overtake frame
relay networks, predicts Rich Glasberg,
manager of data communication for the
Commonwealth of Massachusetts. But that
won't happen for years.
In the meantime, many enterprises will rely on
frame relay for wide-area connections,
including the 350-node, fully meshed network
Glasberg oversees.
Until the day VPNs rule, he wants the forum's
technical committee to be ready. Frame relay
may seem mature today, but unforeseen
challenges could arise that will call for
additional standards.
"I understand they have covered seven-eighths
of the issues, but the truth is the world is still a
moving target. If an issue does come up, who
would handle it?" he says.
While, as a user, Glasberg still sees value in the
forum, the group itself has no end users as
members.
The forum is 69% hardware vendors, 16%
service providers and 15% academics.
Membership has hovered between 135 and
145 companies for the past three years despite
mergers and acquisitions among members,
according to Dreher.
She says the current soul-searching by the
forum is about deciding how to continue
meeting members' needs as well as those of
end users.
"I guess the question is, 'How do we want to
evolve the forum?' We want to give members
value for their membership dollars. But we also
want to make frame relay more meaningful
within user networks the way that they really use
them," Dreher says.
|
