x
Optical/IP

MPLS Gets Lukewarm Reviews

LONDON -- Lightspeed Europe 2001 -- Gurus at the Lightspeed Europe conference questioned the promise of the emerging core network technology standard, Multiprotocol Label Switching (MPLS), during a panel debate held here today.

MPLS proponent Geoff Bennett, who is a distinguished engineer at Marconi PLC (Nasdaq/London: MONI), said he fears that if the industry relies on the Internet Engineering Task Force (IETF) to standardize MPLS, disappointment may follow.

"There may be little to prevent MPLS from going the way of ATM or OSI," he said. "Standards bodies today aren't about making standards, they're about politics... about how vendor A can suppress the initiative of vendor B in order to gain competitive advantage."

The IETF, he said, moves from "one slippery slope to another," enabling vendors to snarl the process in order to stifle competititors' plans. "It's a really, really childish way of running a standards body." And, unfortunately, if allowed to continue it could doom MPLS before it really gets off the ground.

Peter Lothberg, an independent consultant who took the opposing side of the MPLS debate, didn't comment on the IETF. But he said MPLS may not even play a minor role in future IP networks if warring factions cannot agree. "We may end up with nothing," he said.

Bennett and Lothberg faced off during a session titled "MPLS: Just Another Marketing Bandwagon?" Bennett maintained that, while MPLS is still developing, it offers the most efficient way yet to advance quality of service in IP networks, particularly if implemented in the network core. It can support traffic engineering fast restoration, and VPNs (virtual private networks), he said.

MPLS also has been touted as a means of unifying the management of so-called legacy services like ATM (see Équipe: Take the ATM Road to MPLS). And it's under discussion as the basis for delivering Ethernet-based services in metro networks (see MPLS Spurs Metro Ethernet Debate).

Lothberg said MPLS isn't required or desirable in the network core, but that it can run as an application in any IP network -- so long as carriers are willing to pay for the work involved.

He suggested that, depending on a series of variables, it may prove to be more economical for carriers to simply add more circuits to support user demand, instead of implementing MPLS, even at the network edge.

"Carriers need to decide whether to spend more money to bump someone off the network or simply to upgrade their facilities," he said.

The discussion culminated with an informal poll of the audience: About one quarter of those present raised their hands to agree that MPLS is "just another marketing ploy." Roughly the same number raised their hands that it wasn't. About 10 percent of the attendees thought there was a credible alternative to MPLS, and about the same thought MPLS is the only way to go. Perhaps significantly, the large majority of attendees didn't vote at all.

Today's session seems to be yet another manifestation of ongoing worries about the future of MPLS. At another conference in November, a group determined that supporters of the protocol desperately need focus their work in order to maintain momentum (see MPLS: Keeping it Real).

This isn't the first time the IETF's role has been called into question. Even longstanding supporters of the group say it's turned into "a negative force," whether or not MPLS becomes a standard (see The Monster Memo and IETF Routing Director Resigns).

— Mary Jander, Senior Editor, Light Reading
http://www.lightreading.com
Page 1 / 4   >   >>
temporary 12/4/2012 | 11:04:32 PM
re: MPLS Gets Lukewarm Reviews MPLS is less a technology than an agreement
between routers and edges on route management.

Data moving via MPLS must have some agreement at the network edge to maintain QOS. These prior agreements are aggregated to travel assigned to a single MPLS label. The MPLS label represents a secondary QOS agreement between edges and network domains used for transport.

So, I guess logically, the network edge could transfer the entire QOS contract, based upon IP identifiers, into the network domains via a signalling method, using "out of band" signalling.
This involves skipping the intermediate (MPLS).
Then the network routers would look up the proper QOS method using the IP identifier rather than the MPLS label.


Whats wrong here?




temporary 12/4/2012 | 11:04:32 PM
re: MPLS Gets Lukewarm Reviews MPLS is less a technology than an agreement
between routers on route management.

Data moving via MPLS must have some agreement at the network edge to maintain QOS. These prior agreements are aggregated to travel under a single MPLS label. The MPLS label represents a secondary QOS agreement with network domains used for transport.

So, I guess logically, the network adge could transfer the entire QOS contract, based upon IP identifiers, into the network domains via a signalling method, using "out of band" signalling.
This involves skipping the intermediate (MPLS).

Whats wrong here?




mplsrocks 12/4/2012 | 11:04:16 PM
re: MPLS Gets Lukewarm Reviews I think if the vendors just stay focussed on the main purposes of MPLS: traffic engineering, QoS, and IP VPNs then they could do themselves a lot of good...

And in fulfilling these aims of MPLS, the vendors have come up with such 'creative implementations' of MPLS that they sure have scared the carriers away. If they could just concentrate on making MPLS look less like ATM and complementary to IP, then they would bring a lot of smiles to the carrier's face.
ARBoy 12/4/2012 | 11:04:15 PM
re: MPLS Gets Lukewarm Reviews Obviously, you're both inexperienced in both technology and the day to day interface with large carrier customers.
skeptic 12/4/2012 | 7:29:33 PM
re: MPLS Gets Lukewarm Reviews
I personally take the fact that these particular
people are upset as a sign that things are
going better for MPLS.

Some of these people are upset that IETF has
not reacted fast enough to their great plans
to stuff every bit of garbage in ATM into MPLS.
Or they have strange ideas that IP is going to
disappear into an exclusively MPLS world. The
IETF has already gone way too far in giving these
people what they want.

As far as QOS goes, MPLS doesn't really offer
much of anything. I dont know why people think
otherwise.

As far as MPLS goes, the useful parts of it will
be finished when fast reroute is done. Everything
else being discussed is mostly useless and
should be ignored or has nothing directly
to do with MPLS.






(_x_) 12/4/2012 | 7:29:30 PM
re: MPLS Gets Lukewarm Reviews Well put!
WeNoNeedNoStinkinMPLS 12/4/2012 | 7:29:29 PM
re: MPLS Gets Lukewarm Reviews This is the classic debate that has been going on forever. If my memory serves me correctly wasn't it the FORE team that stated very clearly that "ATM was going to be everywhere". What you have now is the same group of people that believe that you couldn't possibly scale a network without having a connection oriented protocol gave up on the notion of IP running OVER ATM and decided "we'll get our circuits somehow" and put together MPLS as a way to put circuits OVER MPLS.

There's a reason why the Internet has scaled to what it has, it's connectionless. No the Internet is not going to solve world hunger and every possible application that every vendor wishes it could. MPLS is NOT the answer. The peer to peer network that we built has political boundaries between them that no protocol will ever overcome. It's not the fault of the IETF or any other standards body.

The Internet is a best effort network. Carriers have not deployed QOS even within their own domains, let alone across domain boundaries. Lets get over all the marketing and come back to Earth.

By the IETF bringing MPLS into the forum, they've effectively collapsed all the losers from the ATM forum and hence the IETF gets stopped up with the same people trying to make the same mistakes over and over again.

We learn history so history doesn't repeat itself...I've seen this movie...I know how it ends... Did you sleep through it?
Titanic Optics 12/4/2012 | 7:29:28 PM
re: MPLS Gets Lukewarm Reviews >>There's a reason why the Internet has scaled to what it has, it's connectionless.<<

Hmmmm...I didn't realize that the Internet was a true datagram, connectionless network.

Is the Internet connectionless on an intercity, long-haul basis? I understood that the backbone (Q, T, BRW, WCOM, CWP, etc.) carriers use ATM over SONET connections in between cities, having mapped the IP packets into ATM. If I am in LA, and I send an IP packet to New York city, what carrier has a true datagram approach for that?
edgecore 12/4/2012 | 7:29:27 PM
re: MPLS Gets Lukewarm Reviews Is not the goal to have MPLS improve the internet experience?

Is it wrong to say QOS in the same breath as MPLS. Is QOS only Diffserv and RSVP?

EC
sesej 12/4/2012 | 7:29:25 PM
re: MPLS Gets Lukewarm Reviews
QoS is just one part of an over all "Policy" based network. QoS is just a priority for any "queue" that a packet may encounter. True policies also include SLA's for latencies, delays, committed bandwidth, etc.

The "convergent" future requires much more then just QoS, it will require policy based networks.

MPLS has little to offer for policy based networking. But I think the answer is already at hand and would be simply using level 4 (application) packet forwarding (maybe in conjuction with RSPV)
Page 1 / 4   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE