Are “Managed Services” a New Thing?

The reaction to the Internet regulation framework Google and Verizon laid out on August 9th by the tech blogs was extremely uniform: the bloggers harshly criticized the firms, Google in particular, in very personal terms (“sellout,” “surrender monkey“, “greedy swindler,” etc.) and lambasted the agreement for its failings in terms of mobile broadband and managed services.

The problems the bloggers had with mobile don’t strike me as very substantial. They know that Google sought and obtained net neutrality restrictions on the C block frequencies in the 700 MHz auction and entered a bid to ensure that the reserve price was met. They must also know that all of the other spectrum auctions didn’t have net neutrality conditions, so it’s not particularly kosher for the government to change the rules after cashing the checks; we call that “bait-and-switch.” The unencumbered spectrum cost three times as much as the net neutrality spectrum, as you would expect. If public policy demands net neutrality restrictions on spectrum, they would need to be imposed before the spectrum is auctioned; but that’s not what this post is about.

I’d like to discuss the other issue that upset the bloggers, the so-called “managed services” exception to the anti-discrimination rule that Google and Verizon propose for wireline Internet services. The bloggers all said that this was a dangerous precedent, or words to that effect. GigaOm’s Stacey Higginbotham said managed services don’t exist yet:

[Managed services] is theoretical today, and is where the potential for big controversy lies.

TechCrunch’s M. G. Siegler said managed services would set a “horrible precedent:”

With the wired aspect, there’s still a portion that would give providers the ability to circumvent this neutrality with their own services (under proposed government supervision). With wireless, it’s more of a Wild West situation. Things are still shaking out, and there’s plenty of competition, is the argument.

Both arguments are faulty because both are extremely gray. And both set a horrible precedent going forward (if adopted, of course).

Ars Technica’s Matthew Lasar scratched his head over managed services:

But what exactly will these “additional or differentiated” services that ISPs could charge content providers extra cash for be? IP Video? The latest, coolest live conferencing app?’s Eliot Buskirk denounced managed services as a yet-to-be-created new, paid Internet:

In other words, to avoid creating tiered access on the internet and dealing with associated governmental red tape, Google and Verizon have proposed creating a second, paid-access-only internet — and mobile networks are exempt from the proposal, so the concept of net neutrality wouldn’t necessarily apply there.

The tech bloggers are in agreement that managed services are some sort of new wrinkle that would be bad for the Internet. Whether they’re good or bad is a long conversation that impacts innovation costs, investment, and the nature of applications and network services. For now, I’d rather stick to the much more factual question of whether managed services are new.

The brief answer is: “No, managed services are nothing new.”

Lasar came closest to grasping the concept when he speculated about “the latest, coolest live conferencing app.” This is a great example, and timely as well.

By way of demonstration, let me fill you in on one of the greatest deals in technical analysis on the Internet, GigaOm Pro. This is a modestly-priced (less than $100/year) subscription service that provides access to professional research that goes well beyond the scope of any of the free information you’re likely to find on the Internet if you don’t scour academic research papers (as I do, alas.) That was a free testimonial.

There’s a report on GigaOm Pro that explains how corporate video conferencing is done*, Videoconferencing Unleashed: The Enterprise Videoconference Landscape, 2010 – 2015, by Lisa Pierce. Pierce describes how enterprises do high-resolution video conferencing in some detail, who the players are, and what sort of services they offer. The quick summary is that enterprise video conferencing doesn’t rely on generic technologies like PCs and the public Internet, but rather on special purpose hardware and software and an Internet bypass that uses IP over MPLS on private lines:

Many high-end videoconferencing technologies, like Cisco’s TelePresence or HP’s Halo, require the use of secure, high-performance WANs. Typically enterprises employ MPLS or high-speed private line services for this purpose. But companies use different carriers for WAN services; with the exception of Internet services, few carriers currently support WAN interoperability for retail (vs. wholesale) purposes.

In the US, these facilities don’t cross domain boundaries very well, so there’s an interoperability issue for conferencing between different enterprises that is being addressed by standards work:

Some carriers have begun to develop provider-specific internetworking agreements to support interoperability for select videoconferencing systems on an international basis. And one international industry group, the i3 Forum, is tackling inter-carrier interoperability to support immersive HD videoconferencing.

While our cousins in Europe are far along the curve on immersive conferencing across network boundaries, this is only done in the U. S., according to Pierce, by service bureaus:

But there is nothing like this for domestic use. In the U.S., for example, there is a lack of MPLS interoperability between AT&T and Verizon to support any retail customer application.

Today, business use video bridging and exchange services to facilitate videoconferencing sessions between companies that employ different carriers, WAN services, etc.

So that’s one of the problems that the managed services exception aims to solve. It’s not a new problem, and it’s not without precedent; people in the U. S. address it today by buying into a video conferencing service that bypasses the Internet and securing a private line (or virtual private line) to access the service. Full interoperability depends on standards, extremely robust networks (much better than you need for email, web surfing, and Netflix streaming) and a legal environment where such services can be offered as both retail and wholesale services.

Let me stress that this is not Skype video calling, this is Cisco TelePresence where you have a wall full of monitors, multiple high-definition cameras, and automatic focus on the current speaker. This app takes a lot of bandwidth and must be lower in latency and jitter than the public Internet, where traffic surges and declines in cycles dictated by TCP’s congestion avoidance system. You can’t support this application on communication channels that also carry a lot of TCP traffic for a reasonable price in a best-efforts only (“all packets are equal”) scenario.

The other two examples of managed services that are currently in widespread use are cable TV services like AT&T U-Verse and the managed VoIP that’s a part of all carriers’ Triple Play packages. There was a carve-out in the AT&T – Bell South merger agreement that permitted managed services and also applied a general open access rule to Internet services, so even that is nothing new.

So the answer to the question we asked in the title of this post is: “No, managed services are nothing new. There are many apps that require a level of service that’s simply not practical on the generic Internet today, but are within the capability of the MPLS networks that carry TCP/IP traffic for ISPs. Managed services is an enabler of these applications that allows leading-edge firms and consumers to deploy them today without affecting the Internet.

I said I wasn’t going to address innovation, but you can connect the dots: a wholesale ban on managed services certainly impedes innovation in some types of applications. I think it’s reasonable to conclude that stifling this type of innovation harms the Internet over the long term by taking away a target toward which to develop, but as I’ve said, that’s a long discussion that bears on the end-to-end arguments, best-effort networking, and a thorough examination of Quality of Service engineering.

At some point, it should become practical to fold these applications over to the generic Internet, but that’s going to take some advances in compression, processing, QoS, and service plans. In the meantime, it seems reasonable to allow the people who want to run these apps to buy the services that make them work. Am I missing something?

(*Footnote: Yes, I do find it interesting that GigaOm is telling one story to the masses who read the site for free and a more nuanced story behind the paywall. I suppose that’s just the way things are on the Internet: the ad-supported content is always going to be more sensationalistic and eyeball-grabbing than the paid content. There’s no free lunch, even on the Internet.)