If the Internet shifts out of neutral, what might that mean for the Enterprise WAN market?

August 16, 2010

Shanghai Nights

During last week’s bombardment of Verizon/Google analysis I found myself  speculating on the possible “death” of net neutrality and the potential implications from my own perspective in the carrier industry. This was one of the questions that popped into my head:

What is the difference between:

  1. A private WAN
  2. A secure VPN on a prioritised/tiered, non-neutral Internet

Those two things could bear a striking resemblance to one another, couldn’t they?

Generally speaking, private wide area IP networks have value to business because they provide what the Internet does not – predictable, reliable performance. They deliver critical enterprise-grade performance in a world where all Internet traffic is treated as best-efforts, leaving everyone at the mercy of congestion and packet loss. If premium, prioritised Internet was on offer for the right price, leading the performance of Internet VPNs to dramatically improve, what would happen to the value proposition of the traditional IP WAN offering?

If you view the Internet in a political sense what it essentially boils down to is a collection of agreements to exchange traffic between networks. Those agreements determine routing policy; decisions on where to send traffic, based partly on efficiency but mostly on economics. It is the bread and butter of every autonomous system to apply policy to control where traffic should be sent to next. While it is technically possible to apply priorities to Internet traffic, in general this doesn’t happen. Most importantly if traffic does happen to be marked as requiring a QoS level this isn’t honoured between networks. So, whilst I concede that some engineering issues may exist, the main obstacle to QoS on the Internet is actually the absence of a common agreement to prioritise traffic marked as “premium”.

You might argue that any kind of meaningful agreement is unlikely, on the grounds that the industry will want to protect its higher-margin enterprise services. I recently stumbled across this fascinating interview with Malcolm Matson, in which he remarked that “the telco industry is probably the largest global cartel that mankind has ever known”. This is fairly accurate. However, if it becomes commonplace for anyone, not just content providers, to have the option to buy themselves out of best-effort Internet then it is feasible that we might see QoS agreements emerging between providers who don’t already play in the WAN market.

Admittedly, there are certain applications that will always demand a dedicated network with an SLA. However if the performance gap between this and “premium” Internet closed significantly, how many IT directors would still buy a global IP WAN?


2 Responses to “If the Internet shifts out of neutral, what might that mean for the Enterprise WAN market?”

  1. Jan Dawson Says:

    Interesting thoughts. I think the bigger challenge is that in shifting from a private WAN to a VPN over the public Internet you shift from a handful of classes of service which are designed to differentiate between traffic owned by a single entity to a model of QoS which has to handle traffic from a multitude of sources.

    I.e. a private WAN might have 6 CoS, but they all apply to an enterprise’s traffic in a way it specifies and has complete control over. Essentially, as long as I never send more priority 1 traffic than I have bandwidth to handle it, it will always get through with the characteristics specified. But on the public Internet, my priority 1 traffic is now just a tiny sliver of total priority 1 traffic, and competes with the rest. And as soon as total priority 1 traffic exceeds total bandwidth available, it gets degraded. Unlike the enterprise using a private WAN, I can’t simply pay for more bandwidth in this scenario, because no one player controls all the necessary links that need to be upgraded and because, in fact, I’m not directly paying for any of them. So there’s a huge loss of control between a private WAN where I control both total bandwidth available and the absolute priority given to my most important traffic, and the public Internet VPN model, where I have no control over total bandwidth and little control over the relative priority given my traffic beyond a certain point.

  2. Dan Young Says:

    Yes I agree, the end user has much less control in this scenario. There are also the contention and over subscription issues.

    However a typical multi-site Internet VPN might only involve a very small number of providers, selected by the customer.

    I think the feasibility of any solution would be dependant on a) The ‘transparency’ of the providers involved (regarding information given to customers)
    b) How complex the inter-AS agreements could reasonably be (when it comes to bandwidth and upgrades)

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: