That future was a thing of the past

By Scott Bradner
Network World, 11/05/01

The demise of Sprint's Integrated On-demand Network service has finally been accepted by all concerned, though it has taken a very long time.

Sprint launched ION in June 1998 with great fanfare, after which it quickly disappeared as far as most of the world was concerned. Apparently Sprint has been busy behind the scenes shoveling money into ION all this time, as much as $5 billion according to the Network World obituary.

I wrote about ION in 1998 when it was first announced and was not too positive about its prospects: "It is far from clear if Sprint will be able to actually make a go of ION . . ." Because ION was based on ATM to the customer, I mostly dismissed the possibility of anything technically viable coming of the effort, but some of the nontechnical aspects were interesting.

Particularly interesting was Sprint's plan to move away from minute-based phone billing to billing based on the amount of data sent, whatever the application. That may still happen, but not because the carrier decides to offer voice service on a per-bit rather than a per-minute basis. It will happen because users will just use their Internet connectivity to replace their current wired phone service and the carrier will not be able to tell what is going on.

I expect it's just a coincidence, but Sprint announced it was giving up on ION a week before Microsoft's introduction of Windows XP, which, with its built-in Session Initiation Protocol (SIP) support, will make this type of call-around much easier.

SIP is the IETF's multimedia signaling protocol, as defined in RFC 2543. SIP is getting a lot of traction in the IP-telephony world these days. In addition to XP, SIP-based Ethernet-connected phones and telephony servers are on the market. SIP has the potential to significantly affect the traditional phone world by enabling individuals and companies to easily bypass telephone service providers.

But I don't think the root cause of the ION failure was the threat of SIP. I think it was mindset. Specifically, it was telephone company mindset.

In the case of ION, this mindset manifested itself as services based on circuits - ATM virtual circuits to be specific. The Internet has proven again and again that application-specific circuits are not needed and just add technical and managerial complexity, and thus cost. A few million people running SIP-based voice applications will show this yet again. I wonder if anyone in telco-ville is listening.

Disclaimer: Managerial complexity at Harvard? Say it's not so! Anyway, the above prediction of the past is mine alone.