Ethernet services

Counting the Cost of Ethernet Services

Ethernet services are in the cards for most service providers, but the cost of add-ons such as QOS and managed protection is a big issue, according to the "2004 Survey of Ethernet Service Providers" published by Heavy Reading, the paid research arm of Light Reading.

The report, released today, examines carriers' aggregated expectations for Ethernet services, grilling them about the types of services to be offered and the carriers' attitudes toward the perceived pros and cons of Ethernet. The report also correlates the findings with those of a July survey that examined Ethernet services from the enterprise point of view.

Most respondents said their companies plan to offer Ethernet services by the end of next year, if they don't already, and it's clear that they see Ethernet growing at the expense of traditional Layer 2 services such as Frame Relay and ATM. The economics of those services are far from settled, however.

Big questions arise concerning quality of service (QOS) and managed protection, two features associated with traditional telco services. These don't come for free in Ethernet -- and yet, users might demand them for free.

In July, a Heavy Reading survey of enterprises found that 31.6 percent of respondents believed Ethernet QOS shouldn't cost extra. Another 38.5 percent said they wouldn't pay more than a 10 percent premium for QOS. Managed protection services got even less love. The July survey found 71.8 percent of respondents expecting to pay zero for managed protection, in some cases (12.9 percent of all responses) because they just didn't need it.

So, carriers might have to pony up the costs of QOS and protection without getting more revenues in return. The punch line is that those extra costs remain undetermined. In the survey, 34.1 percent of carrier respondents said they didn't know what it would cost to provide QOS for Ethernet services, and 44.5 percent said the same for managed protection on Ethernet private lines.

Other carriers seem to think they'll be able to recoup the costs with premium prices -- an option that contradicts the enterprise survey. For example, 26.2 percent of carriers surveyed said QOS would increase the cost of Ethernet services by 10 percent to 25 percent. And 25.2 percent believed the price to customers could be raised by a like amount if QOS were added.

— Craig Matsumoto, Senior Editor, Light Reading

For more on this topic, check out:

For further education, visit the archives of related Light Reading Webinars:

yakatori_guy 12/5/2012 | 1:11:05 AM
re: Counting the Cost of Ethernet Services ...and it is us.

So end users want something for nothing? Unfortunately, IMHO we in the Ethernet community have done a great deal to set that expectation. In our euphoria to promote Ethernet services, we have trained end users to expect a free lunch...services just like private line or Frame Relay, but at significantly less cost per unit of bandwidth. All because Ethernet equipment is so inexpensive to install and operate...and beneficent service providers will pass those savings on to end users.

Truth is, even simple Ethernet networks cost real money to install and operate (and never mind to scale). Furthermore, Ethernet networks that provide similar service attributes as private lines (or even Frame Relay), such as guaranteed BW, QoS, and protection, can cost even more.

I'm all for Ethernet services -- it's part of my job to promote the Ethernet equipment that my company sells. Ethernet services can offer unique benefits to end users. I just think that, as an industry, we have done a disservice to Ethernet by setting unrealistic expectations with end users. Hopefully, that will not come back to bite us...

sigint 12/5/2012 | 1:11:02 AM
re: Counting the Cost of Ethernet Services yakatori_guy:
Ethernet services can offer unique benefits to end users.

Unique services other than cost? Could you elaborate?

Excellent observations in your post, BTW.
Frank 12/5/2012 | 1:11:00 AM
re: Counting the Cost of Ethernet Services YouGÇÖve made some very good points, YG. But those points are mitigated, IMO, by the pricing trends taking place elsewhere in the same space, irrespective of what users perceive as a free lunch model relating to Ethernet.

All infrastructure costs, starting with semis and extending to the costs of lasers and optical ports and beyond, are plummeting from a cost per bit perspective, so the incremental cost penalties associated with improved quality and reliability viewed against a backdrop of Ethernet's propensity to spread ever wider, which only adds to its economies, must be taken with of view of those factors, as well. Unless, of course, someone wishes to posit that this phenomenon has in some way bottomed out or plateaued at this point, which I truly doubt.

You are correct in noting that it is more expensive to include those features, however, esepcially when they are initially introduced. And I would add that it is even more expensive to administer Ethernet outside the LAN due to the failure of manufacturers to include OAMP and surveillance capabilities into SFP/XFP modules, a factor that causes even the staunchest IP proponents to favor SONET where rock solid network operations are concerned.

But the latter shortcomings are not instrinsically the faults found within Ethernet _as _much _as they reflect the intent or lack thereof behind what the standards bodies and form factor associations deem necessary to satisfy their needs, and not always necessarily the needs of service providers.

But those incremental costs are at the same time taking place on a parallel declining slope with that of plain vanilla Ethernet, not quite at the same rate, granted, but declining just the same. Maybe equating to an asymptotic trend, similar to how the cost of a DTMF touch-tone pad is no longer being considered an added luxury feature when provisioned on a kitchen wall phone.

Frank Coluccio
yakatori_guy 12/5/2012 | 1:10:54 AM
re: Counting the Cost of Ethernet Services yakatori_guy:
Ethernet services can offer unique benefits to end users.

Unique services other than cost? Could you elaborate?

Some potential advantages over private line/Frame Relay:

1) Bandwidth -- Can scale up to 1 Gbps, and even 10 Gbps. Many services scale in fairly granular increments (e.g., 1 Mbps or 5 Mbps).

2) Flexibility -- It is possible to adjust the service bandwidth through provisioning. Some providers even allow the customer to do this using a web-based interface. This eliminates the cost and time associated with scaling, for example, from a single DS1 to two DS1s.

3) Emulated LAN capabilities -- E-LAN Ethernet services can create an Ethernet Layer 2 VPN -- basically the service provider network emulates a LAN. This can result in lower costs for the end user (depending on the application).

4) Lower CPE costs -- Ethernet interfaces tend to be significantly cheaper, per unit of bandwidth, than DS-n or OC-N interfaces on CPE routers or switches. Saves the end customer a few bucks.

5) Data friendly -- IT folks tend to know Ethernet much better than TDM. In theory can lower customer opex; really probably helps decrease objections to making the sale.
douggreen 12/5/2012 | 1:10:33 AM
re: Counting the Cost of Ethernet Services yakatori_guy

In my experience from selling similar kinds of function in the real world, some of the "benefits" you mention are difficult to take advantage of:

1. Scale in granular increments. An infinite amount of granularity is impossible to price in a way that works for both customers and providers, and produces a nightmare for network planners and operational support systems.

2.Flexibility to adjust bandwidth: Translated: I get to deploy 10 gig of bandwidth to the user, provide over-capacity in the network, and deploy overly-expensive CPE just in case the customer might want that amount of bandwidth some day, but I only get to bill them for the 10 meg they want today. Once again, a planning and financial nightmare. I do admit that there are some premium services that might charge more up front to allow this capability in a limited manner, but this is not the "cheap" bread and butter service that Ethernet promises to provide.(FYI, SONET vendors are starting to include this kind of function as well, but they face the same issues in network planning and up-front cost.)

3. Emulated LAN capabilities.. I have that today with ATM, Frame Relay, and even using an IP network with TDM as an access method. Just stick a VPN router on the end. Yes, you may be able to do it somewhat better with Ethernet, but this alone is not enough. (FYI, your sales pitch could have been copied verbatim from one I used 10 years ago to sell ATM based Ethernet LAN services)

4. Ethernet CPE is cheaper: THe cost difference is usually associated with three things: 1)the OAM and reliability features that the customer want added to Ethernet 2) Historically expensive and over-engineered SONET equipment. This started to change with Fujitsu and Cerent, and has being pushed in the extreme by companies like White Rock 3)Higher margins on SONET equipment that can and will go away if the SONET vendors feel threatened enough by Ethernet. Any other cost differences are minimal. There is no reason that similar function needs to cost more.

5.Data Friendly to the customer: True, but once again, the customer facing side of the CPE can be Ethernet without changing the infrastructure. What you describe is already the case with ATM, managed router and managed Ethernet CPE services. Several companies are specializing in making Ethernet services work well over existing TDM networks (see Overture Networks for an example).

The devil is always in the details.

Sign In