MPLS Is Growing Up

Multiprotocol Label Switching (MPLS) has beaten out Asynchronous Transfer Mode (ATM) for the carrier core, and new carrier deployments are rolling out as vendors shift their focus from getting the technology to work, to figuring out how to make money from it. “MPLS is no longer a pie-in-the-sky technology,” says Steve Vogelsang, vice president of marketing at Laurel Networks Inc. “The technology has matured to the point where we’re not talking about what cool new widget has just come out. We’re looking at how we can actually manage these networks.”

MPLS's maturity was apparent at the three-day MPLS 2003 conference in Washington, D.C. For six years the conference, organized by testing lab Isocore, has attracted the most influential MPLS engineers. Conference organizers say more than 500 registered this year, compared to last year's attendance of 350. Isocore's Bijan Jabbari says the increase was due to more service provider participation. For the first time, Bell operators took part. BellSouth Corp. (NYSE: BLS) even co-chaired the event, while Verizon Communications Inc. (NYSE: VZ) sent a representative to speak on a panel.

BellSouth, AT&T Corp. (NYSE: T), MCI (Nasdaq: MCIT), and NTT Communications Corp., among others, have already started deploying MPLS. Others, including Verizon, are in the process of selecting gear and deploying it. Even Sprint Corp. (NYSE: FON) seems to be coming around to the idea of using MPLS in certain parts of its network.

From a technical perspective, a lot has happened since the last MPLS conference in October 2002. Standards work moves forward on MPLS fast reroute, a mechanism that sets up alternative paths for sub-50ms recovery and restoration. Isocore demonstrated interoperability of the technology at the MPLS 2002 conference (see MPLS Vendors Demo Fast Reroute). Layer 3 virtual private networks (VPNs) based on the Internet Engineering Task Force (IETF) proposal 2547 also continued to grow in popularity among carriers (see MPLS VPNs: The Talk of Supercomm). Carriers are offering point-to-point Layer 2 MPLS VPN services using Draft Martini as well.

Virtual private LAN service (VPLS), the Layer 2 multipoint implementation of Layer 2 VPNs over Ethernet, has also made significant headway in the past year (see VPLS Standard Debated). The jury is still out in terms of the signaling protocols that will be used to provision the nodes, but a common standards document should be ready by the end of 2004 (see Kompella Backs BGP).

While much has been accomplished, there's still more work to do. Here’s a look at some of the most pressing issues discussed at this year’s MPLS conference:

  • Intercarrier MPLS
    Carriers have mastered the technical intricacies of offering MPLS VPNs and traffic management within a single carrier network. Now the challenge is connecting those networks to others so that end-to-end quality of service can be guaranteed. Billing issues and management between carrier domains must also be addressed.

    “MPLS is designed to be effective in a single-carrier network,” says Christian Martin, a Distinguished Member of Verizon’s engineering team. “But you can’t do traffic engineering between carriers. There are some techniques proposed, but there’s more work to be done here.”

  • Operations, Administration, Maintenance, and Provisioning
    Now that an MPLS network has been deployed, how do you manage and troubleshoot problems? ATM and Frame Relay already have trusted OAM&P mechanisms. Carriers, especially the regional Bell operators, want it for MPLS, too.

    “MPLS provides very basic monitoring and only tells you if you’ve lost a path,” says Peter Hill, vice president of technology planning and development at BellSouth. “But it won’t give you any information about what happened on that path.”

    Some vendors say that a separate OAM&P mechanism isn’t needed. Vogelsang of Laurel says that troubleshooting can be done using existing IP routing protocols.

    “OAM&P is a big deal for service providers,” he says. “But there are other mechanisms and protocols that can tell you what’s happening on the network. Of course, if that’s what customers want, we’ll give it to them.”

    The big question mark now is which organization will develop the standards. Will it be the IETF or the International Telecommunication Union (ITU)? Both groups are currently working on solutions.

  • Multicast Over MPLS
    Unlike other technologies like ATM or IP, MPLS doesn’t support a one-to-many communication method. Adding multicast capabilities would allow carriers to offer new services like video on demand for residential users. NTT has added extensions to MPLS and has been experimenting with MPLS multicast. Representatives from the carrier presented information at the MPLS 2003 conference. Juniper Networks Inc. (Nasdaq: JNPR) also claims to offer a solution.

  • Generalized MPLS
    Last but not least, engineers will continue working on integrating IP and optical signaling over MPLS. Even though the standards are still a long way off, and carriers don’t seem terribly interested in the technology just yet, vendors are already working on solutions. Isocore has already done an interoperability test of some of the GMPLS gear that’s now available. (For more on this, see today's story: GMPLS Showcased in Demo.)
— Marguerite Reardon, Senior Editor, Light Reading

Page 1 / 3   >   >>
boozoo 12/4/2012 | 11:18:17 PM
re: MPLS Is Growing Up A minor error in the article: Video on demand, unlike pay-per-view is a service implemented using unicast technology rather than multicast.

right_leading 12/4/2012 | 11:18:16 PM
re: MPLS Is Growing Up
gea 12/4/2012 | 11:18:15 PM
re: MPLS Is Growing Up "Multiprotocol Label Switching (MPLS) has beaten out Asynchronous Transfer Mode (ATM) for the carrier core,"

If you mean that there's more MPLS in the core than ATM, I would highly, highly doubt that.

If you mean that most big carriers are now planning to build out their cores (eventually) with MPLS, even that I'd bet is an overstatement.

OAM&P rules the big networks. If they can't manage it, they won't deploy it. If MPLS does not give ILECs, etc...a 'nice' (meaning SONET&ATM-like) way to do OAM&P, then it'll die a slow death in the US.

Fortunately, I think there's enough momentum and interest to make MPLS OAM&P standards come around pretty quickly.
Dr.Q 12/4/2012 | 11:18:14 PM
re: MPLS Is Growing Up Boozoo wrote: "Video on demand, unlike pay-per-view is a service implemented using unicast technology rather than multicast."

Being a hardware guy I'm only making educated guesses here, but it seems to me

a) that a system would do pay-per-view with unicast technology and a simple meter in the CATV descrambler on top of your TV set. The meter logs that you are watching the premium content and relays the information to the billing clerks. (Simple analogy would be using telephone audio line to dial in to listen to a press conference or sports event--the system measures that you're using the service, but everyone gets the same content.)

b) that video-on-demand requires multicast technology if you have more than one viewer demanding content. A separate transmission is required for each viewer. (Simple analogy is a website serving multiple users simultaneously. Each user can get different information from the site in response to their own queries. This is limited by hardware in two ways -- number of ports available and processing bandwidth.)

I would appreciate comments & explanation from folk versed in how the systems REALLY work, or links to good websites that explain the technology.

- Dr. Q
signmeup 12/4/2012 | 11:18:14 PM
re: MPLS Is Growing Up I'm not sure I would go that far...

If by being right means that you developed a product before the market was ready, ran out of money, and had to be bought for basically scrap, then yes, your definition would be correct.

Personally this entire discussion is moot until better management tools come along for MPLS.

JackRJ45 12/4/2012 | 11:18:13 PM
re: MPLS Is Growing Up Extreme Networks makes a product that does Unicast Replication. This gives two major benefits, reduces workload from the servers and allows streaming audio and video to be transported over different AS/provider backbones. Interprovider MPLS was one of the points cited in the article, and subsequently interprovider multicast is the same issue.


Marguerite Reardon 12/4/2012 | 11:18:13 PM
re: MPLS Is Growing Up This has been corrected in the story.
IP Everywhere 12/4/2012 | 11:18:13 PM
re: MPLS Is Growing Up I can't agree more with either comment. Boxes like Tenor, corona, tahoe, celox, etc. were ahead of their time and in a market that was completely disinterested with any startup and where VCs were running for the hills from telecom investments. They worked very hard at making a fancy GUI to try to solve the rapid service delivery problem. The GUIs were good but at the end of the day the GUI still did not matter unless the startups were selling to a CLEC which would not have all the OSS integration issues which come with ILECs. Without a big brother none of the those companies would survive the death march of headcount reduction after headcount reduction.
edgehead 12/4/2012 | 11:18:12 PM
re: MPLS Is Growing Up Glad to know MPLS is mature - I still don't see carriers making much off of it yet until they can provide enterprises with some compelling quantifiable reasons (opex dollars and productivity gains)to migrate away from Frame Relay.-á MPLS VPNs are just a site to site connectivity mechanism that in may ways lacks the ability to address core enterprise WAN issues.-á Enterprises are consumed with security concerns running over an IP infrastructure and how to take advantage to the revolution that is occurring in the BB remote access arena.-á Until Cisco and Juniper realize this, they're going to continue to lose business to vendors that
deliver a portfolio services that address a broader and compelling set of enterprise issues.
strands555 12/4/2012 | 11:18:11 PM
re: MPLS Is Growing Up from the article: "Adding multicast capabilities would allow carriers to offer new services like video on demand for residential users."

If "carriers" deploy TDMA PON, multicast has no significance or value. PON is a PHYSICAL broadcast architecture. Every packet goes to every endpoint on the other side of the dumb splitter, no matter how you label it.

Also, I don't know how the story used to read, but the above excerpt is still not correct. Multicast would only be used for NVoD (near video on demand), where a program has multiple pre-set start times per hour. True VoD starts the instant a user selects it. It's highly unlikely another user would pick the same thing at exactly the same time. If they pick the same thing, except 1 minute later, it still cannot be multicast because of the time skew.
Page 1 / 3   >   >>
Sign In