x
Optical/IP

The New Legacy Network

Remember Asynchronous Transfer Mode (ATM), the networking breakthrough of the 1980s? You know, the voice crying in the wilderness, preparing the way for the Kingdom of Internet Protocol (IP) on carrier networks?

Things haven't gone quite as planned. Turns out that ATM, not IP, forms the basis of one of the telecommunications industry's hottest "new" equipment categories.

According to "Retro Chic: Multiservice Switches," the latest report from the Optical Oracle, Light Reading's paid subscription research service, carriers not only like ATM, they're expanding its use.

Based on in-depth interviews with AT&T Corp. (NYSE: T), BellSouth Corp. (NYSE: BLS), and Verizon Communications Inc. (NYSE: VZ), the report finds that ATM's endurance is due to the fact that it makes money.

One of the dirty little secrets of the networking industry, the report finds, is that IP services lose money. It's data services running over existing ATM and Frame Relay networks that are currently paying the bills for many carriers.

Accordingly, products based on ATM are back in vogue. But these aren't your father's ATM switches. Instead, they're substantially improved models (or entirely new designs) featuring greatly improved ATM scaleability and a migration path to next-generation IP networks and services via Multiprotocol Label Switching (MPLS).

Multiservice switch vendors say their approach protects carriers' investments in their existing infrastructures, while reducing the risks involved in building out those networks. Service providers get to hang onto the ATM infrastructure they know and love (and which their staffs are trained to support) and make incremental improvements to it. At the same time, MPLS allows them to future-proof their networks and roll out new services.

It's supposedly the best of both worlds. But the service provider interviews in the Optical Oracle toss some cold water on several pleasant assumptions that have become conventional industry wisdom (if anything in this industry can be said to resemble wisdom, except in the loosest possible sense of that term).

For one thing, it's clear that MPLS is stalled, or at least delayed. Despite what vendors say, most carriers are not ready to deploy it. "MPLS is a longer-range objective," says Larry Tiedt director of broadband switching and routing at Verizon. Others echo this thinking: "When MPLS proves itself, we'll see mission-critical customers consider it," says Trent Long, ATM national product manager at AT&T.

If it turns out that enough service providers drag their heels on installing MPLS, either for capex reasons or concerns over MPLS's immaturity, the decision of some vendors to standardize on packet-based architectures could backfire.

Further, it seems carriers will resist buying from startups. Two of the three carriers interviewed (AT&T and BellSouth) said they wouldn't consider buying from startups at all. "The viability of some of these companies is in question," says Chris Noll, research director for science and technology at BellSouth. "When you get down to choosing a product, it may not be the best can opener." AT&T completely rejects the idea of going with a startup; only Verizon seems open.

Clearly, a competitive market's shaping up. And as usual, vendors are playing games with product specifics. One vendor exaggerated its capacity by a factor of 40. Other claims center on the relative merits of cell-, packet-, and bit-based architectures (all outlined in the report). Also in question is provision for IP routing, which isn't even supported by some multiservice switch vendors.

The report, compiled over three months, profiles nearly two dozen products from the following companies:



— Mary Jander, Senior Editor, Light Reading
http://www.lightreading.com
"Retro Chic: Multiservice Switches" is available for $400 as a single report, or included as part of an annual subscription to Optical Oracle. To subscribe to the service or purchase a report, go here.

Want to know more? The big cheeses of the optical networking industry will be discussing this very topic at Opticon 2002, Light Reading’s annual conference, being held in San Jose, California, August 19-22. Check it out at Opticon 2002.

Register now and save $500 off the registration fee. Just use the VIP Code C2PT1LHT on your registration form, and deduct $500 from the published conference fee. It's that simple!



Editor's Note: Light Reading is not affiliated with Oracle Corporation.
Page 1 / 9   >   >>
AS 12/4/2012 | 10:03:17 PM
re: The New Legacy Network One of the dirty little secrets of the networking industry, the report finds, is that IP services lose money. It's data services running over existing ATM and Frame Relay networks that are currently paying the bills for many carriers
*************************************************

So what exactly are these "IP services" that are not making money. Who has deployed them. Isn't most of the data traffic running on ATM and FR networks IP already.

The carriers are very slow to deploy any new technology. Pure IP based networks without need for ATM and FR switching is still a part of emerging technologies. ATM and FR are already matured technologies and well entrenched. Things are likely to be different in another 5 to 10 years time. Hardly any major carrier can attribute their profitability to data services. The fact is carriers are still to figure out way to make money on data. And continuing to deploy more ATM and FR gear is not going to make it any more profitable for them.
jamesbond 12/4/2012 | 10:03:16 PM
re: The New Legacy Network So what exactly are these "IP services" that are not making money. Who has deployed them. Isn't most of the data traffic running on ATM and FR networks IP already.
------------------

I have heard the term "IP services" millions
of times. I just cannot figure out what the
heck does it mean???????

To me there is only one IP service and that
is "Internet connection". Carriers provide
connectivity to the Internet. Only other
thing I can think of is maybe firewall
type application that could be provided
by the service provider. what else?

Also I think the most common interface in use
for IP would PoS (packet over SONET) and
T1 no?

gigeguy 12/4/2012 | 10:03:14 PM
re: The New Legacy Network "For one thing, it's clear that MPLS is stalled, or at least delayed. Despite what vendors say, most carriers are not ready to deploy it. "MPLS is a longer-range objective," says Larry Tiedt director of broadband switching and routing at Verizon. Others echo this thinking: "When MPLS proves itself, we'll see mission-critical customers consider it," says Trent Long, ATM national product manager at AT&T."

That's a pretty funny quote from AT&T, especially since they've got one of the largest deployed MPLS networks in the world - their "IP-enabled Frame Relay" network. Or perhaps he doesn't read AT&T's own press releases, such as http://www.mplsworld.com/archi... .

For some real facts on MPLS deployments, see http://www.mplsworld.com/archi... , http://www.cellstream.com/MPLS... , and http://www.mplsrc.com/faq3.sht... .
dietaryfiber 12/4/2012 | 10:03:13 PM
re: The New Legacy Network
I think if you looked at the actual FR network that was being run at AT&T (not the press releases), you might find a Router in the middle of their ATM switching network.

Also, back to the IP services....I think IP VPN and Internet Access Services qualify in this category. I suspect that losses associated with this service are based on the low price per bit that is charged with these service. Ask yourself this question: Why should AOL double the size of its network backbone, if the customers using it pay the same amount per month.

dietary fiber
boozoo 12/4/2012 | 10:03:11 PM
re: The New Legacy Network "I have heard the term "IP services" millions
of times. I just cannot figure out what the
heck does it mean??????? "

_________

Here's my 2 cents (trivialized, I'm sure things are more complicated that what follows):

I like to categorize the service offered by an SP (or ISP) as follows (incomplete, just as an example):
- L1 services (T1, T3, lambda etc). Strictly P2P. You need a separate wire/fiber for each destination. You have permanent (leased line) or on demand (ISDN PRI). The customer manages the WAN (implements the routing, is responsible for traffic engineering, etc). SLA tends to be inflexible:
- usually service availability is very high
- QoS very good
- Price tends to be high because of lack of statistical muxing.

- legacy L2 services (FR, ATM). You only have one wire (unless you need more bandwidth). Each destination is represented on the wire by a channel ID (dlci/vci). The SP knows to switch based on the channel ID. You can also have permanent ( e.g. PVC, SPVC) or on demand (e.g. SVC).
Customer manages it's own network.
SLA more flexible, hence you can accomodate different wallets:
- from high availability to no guarantees
- from high QoS to no QoS

==================
And now, the IP services:
==================

- "next-gen" L2 services - (e.g. Kompella) over MPLS. Designed to replace legacy L2 services. Some vendors like to include these services in the "IP services" category, probably because IP is in fashion today. IMO, this is not an IP service.
Major attraction points:
- that they use MPLS (which uses IP routing) which is pictured as the "unifying protocol"
- By introducing the concept of VPN, they promise a certain degree of automation in provisioning which could lead to potential OPEX savings for the SPs which could theoretically translate to cheaper service offerings.
- there's also a flavor of L2 service that does not scale very well called "transparent LAN service". This service extends the customer's ethernet LAN over the WAN. The advantage of this is that it's plug-and-play for the customer, since it provides MAC address learning. This model allow a customer to fully outsource the network from the SP.
- I'm sure there are other pros
- I'm sure there are other cons


- L3 services - (The only one I know is RFC 2547). These are IMO the true IP services.
The major selling point of these services is that the customer gets from the SP an interface that understands IP and can automaticaly route your packets to the right destinations. Unlike for L2 services, where the customer has to statically map IP prefixes to L2 channel identifiers (except TLS).
- In other words, L3 services allows for the customer to fully outsource their datacomm needs from the SPs.
- I'm sure there are other pros
- I'm sure there are other cons

So is a L3 service a more compelling option? I think the answer is: it depends.

For a customer that currently buys a L2 legacy service from SP and has all the infrastructure and expertise in place to run it, the question is: why bother changing to a "next gen" L2 or L3? Unless it's much cheaper.
Let's say it's much cheaper.
Some customers will chose next gen L2.
My argument here is mostly filosofical: the less changes to the infrastructure, the better, because change almost always leads to problems. However, the savings may not be dramatic.

There will be some (Cisco mentality) which will say: outsource, buy a L3 service, fire 50% of the IT folks, and redo the network from scratch because it's going to be cheaper in the long run. We'll feel the pain of change, but it's worth it.

hope it helps,

Boozoo.


Litewave 12/4/2012 | 10:03:11 PM
re: The New Legacy Network One of the dirty little secrets of the networking industry, the report finds, is that IP services lose money.

Regardless of how one defines IP Services, the biggest culprit to it loosing money? Cisco of course!

The TCO of a Cisco Powered Network is obscene.

PS: Isn't it ironic they're featured as the sponsored links after the article :) Nice one LR.
optical_man 12/4/2012 | 10:03:10 PM
re: The New Legacy Network Ok, I understand a couple things from this article:
1) ATM has a place
2) IP has a place
3) LightReading's "Paid Subscription Service" has just released an article stating that ATM is coming back into fashion.

What I gather from that is that there is no new news and that to drum up some revenue LR is stirring up the ATM vs. IP religion wars to make a buck by getting people interested in subscribing to LR's paid report service by messing with the next gen IP vendors minds.

Can we move on?
Thank you.
aran 12/4/2012 | 10:03:10 PM
re: The New Legacy Network In similar context:
Although not exactly state-of-the-art, IBM mainframe sales continue to grow due to the fact that organizations are still heavily invested in legacy equipment. As well, the IBM MF's redundancy and stability cannot be matched by anything else in the market. Technological improvements and extensions to the IBM legacy platform enable the existing customer base to extend traditional mainframe offerings for use with new technologies such as web applications and so forth.
skeptic 12/4/2012 | 10:03:09 PM
re: The New Legacy Network
Don't you think your article is going to be
skewed by the fact that you were talking to
two RBOCs and people at ATT that
are so isolated in their own company that they
don't know whats going on with MPLS at ATT?

I mean do you think the "ATM NATIONAL PRODUCT
MANAGER" is really the most objective person
to get a read on where the company as a whole
is and is going? Did you think you were going
to get him to say that the ATM business isn't
any good or that ATM isn't needed?


indsavvy 12/4/2012 | 10:03:08 PM
re: The New Legacy Network skeptic ---- it's well known that ATM is not going away and that MPLS is still a wannabe technology. If you have any experience of working with carriers, you would know that their adoption of anything new takes many, many years. There's no doubt that the RBOC and ATTs of the world have been playing with MPLS but that's as far as it goes - I know that from firsthand experience. Get real, they're trying to survive in this market.

You seem agitated at the LR article and I can only suppose that you're heavily into the IP and MPLS side of things??? Has it struck home that ATM is the established technology of choice and isn't going to be magically replaced anytime soon - was it a real shock to you? If so, what river have you been paddling up this past two years?
Page 1 / 9   >   >>
HOME
Sign In
SEARCH
CLOSE
MORE
CLOSE