Speaking Engagements & Private Workshops - Get Dean Bubley to present or chair your event

Need an experienced, provocative & influential telecoms keynote speaker, moderator/chair or workshop facilitator?
To see recent presentations, and discuss Dean Bubley's appearance at a specific event, click here

Wednesday, December 14, 2005

IP interconnect: carriers seem to have lost the plot

There's been a lot of noise about the possibility of large carriers attempting to enforce interconnect and policy management regimes on Internet-based services since I last mentioned it. James Enck has been following it closely, and last week at Cisco's Analyst Summit the company tried to make a virtue of it, fitting it around the company's earlier acquisition of P-Cube, which made "deep packet inspection" gear.

During the conference Q&A I tackled SVP Mike Volpi on the topic, and was given a remarkably "scared and blinkered carrier"-centric answer which can roughly be summarised as "broadband is not a human right" and "carriers have to protect their own services on their own pipes, and not have them risked by other people who don't pay them".

Now, I've often seen equipment vendors shrug at their customers' sillier ideas and imply "well, it's their money. If they really want to waste it on something we think won't work, they might as well waste it with us, rather than our competition". However, there wasn't much public eyebrow-raising equivocation of this type. They seem to be serious.

Now, from my perspective there are indeed a couple of reasonable angles to this. Carrier IP networks that are strangled by huge volumes of (usually illegal) P2P traffic are an issue, especially in instances where this could actually break the network. Cellular networks may be especially fragile (imagine a Symbian-born virus which spread using Bittorrent-over-3G) and so I'll buy the notion of packet inspection as a form of pressure valve. Arguably, content filtering should be done in the network too, although personally I think it should be device/terminal-based wherever possible.

But the notion that Internet "service providers" (actually, I see them more as "application providers") like Google or Skype could be forced to pay interconnect fees to carriers (or do co-marketing revenue-share type agreements, as Cisco's Volpi also suggested) is palpable nonsense. The notion that a carrier could explicitly or implicitly degrade Skype of Yahoo!, on the basis that it competes with some of their own services, is horrific to most Internet users, and is likely to stifle the fundamental innovative nature of the Internet.

I struggle to think of a single "cool" and "useful" application or service on the web that was originally developed by a large carrier. They've all come from 2 guys in a dorm at Stanford or 2 ex-Apple employees. Can you imagine the inventors of the next Skype or Hotmail having to negotiate with 327 different carriers and ISPs around the world to make sure their service works properly?

As I said, if IP interconnect is enacted the way the carriers are threatening (largely in the US), it will kill Internet innovation.

Luckily, it won't work. The whole stupid project will fail at several levels - and possibly take the more laudable aspects like P2P traffic-moderating down with it, as it sinks.

The main points of failure:

- All this is scuppered by putting the traffic into a VPN tunnel, originating on a PC or a smartphone. You try filtering something that's encrypted, application-by-application, or URL by URL. Google's supposed WiFi secure access client seems like a poorly-disguised beta for a future version of this.
- In the future, we might build apps out of components more. I reckon XML/.NET based apps and services will be nigh-on impossible to police with deep packet inspection. The same component going across the network could be part of 20 different applications.
- No carriers have a position of "Network Policy Manager", and the complexity of such as putative role in liaising with different departments (network, wholesale, marketing, regulation, legal, HR, large accounts etc etc) makes it unlikely that one will evolve any time fast
- Many apps will just evolve to evade IP interconnect. Someone will make MSN appear to the network as MMS, or Skype will spoof SAP.
- Many operators trying heavy-handed approaches to IP Interconnect will screw up uninintentionally. I'd love to be the fly on the wall when the CIO of a major investment bank rings his carrier account manager and says "I've spent $5m designing a class-leading realtime P2P application to distribute equity derivative pricing information between my traders' desktops and mobile devices. Your stupid network appears to have started blocking it & we've lost $100m through delayed trading. I'm taking my 10,000 users elsewhere, and you'll be hearing from my lawyers and the SEC in the next 30 minutes"

To close off - an open request to the software community. Can you build me a desktop application that will work out (proveably) if my broadband carrier is degrading any particular services? I'll pay you for it. And I'll use it to churn to a carrier that makes a virtue of letting me use my pipes the way I want.

No comments: