Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Ascend Communications-News Only!!! (ASND) -- Ignore unavailable to you. Want to Upgrade?

To: Maverick who wrote (1153)	2/9/1998 7:46:00 PM
From: blankmind	Read Replies (1) \| Respond to of 1629

There's more to IP switching than switched IP I P switching is dead. Nokia bought Ipsilon, the architect of the concept, for a pittance. Can Nokia now redeem the idea? That's about as likely as snow falling in Honolulu. So forget IP switching - it's D . . . E . . . A . . . D. The question now is whether switched IP will die along with it. I say no, but since Ipsilon's apparent success clearly was a victory of hype and hope over substance, I've got to prove my point. IP switching was Ipsilon's term. It describes a way to use ATM to build IP networks that map every persistent user IP traffic stream to its own virtual circuit. That's the problem - every flow gets a virtual circuit. The other switched IP strategies from vendors such as Ascend, Cisco and Toshiba don't work that way - and therefore probably wont follow Ipsilon's IP switching into the grave. The technical problem that IP switching addresses starts with deficiencies in connectionless networking. With router/IP networks of old, all users shared network resources equally and also shared a single address space. Routed IP networks such as the Internet can't exploit increased buyer willingness to pay for services and can't prevent users from freely communicating. Populism such as this flies in the face of reality: Some users have more money to spend, and some want to communicate only with a select subset of the worlds IP population. Service providers are only too happy to take additional user bucks for improved performance and for supporting virtual private networks (VPN) that work like small independent networks inside a public IP network. The fact that buyers are willing to pay for something that providers can't offer creates a major opportunity for some enterprising organization. Where Ipsilon's IP switching went wrong was in failing to recognize that quality counts in networking to the extent that money can change hands to obtain it. Per-flow virtual circuits would have committed service providers to supply private connections for millions of users who never would have paid for them. Furthermore, IP switching used the same address management strategy as routers, a uniform public address space. For non-IP users, or for users who adopt RFC 1918 private addresses to gain flexibility in their own networks, Ipsilon had no answer. Today's virtual circuit networks - particularly frame relay - show that solutions to performance management and address isolation are possible. Almost half of the frame relay services sold in 1997 were focused on building VPNs. Link access routers with frame relay services and you have corporate IP VPNs. The only problem is there is too much per-user cost; buyers report the cost of typical managed frame/router networks as nearly double that of basic frame relay. We need to lower the cost by relying less on the IP device overlay - that is, by bringing IP services into the switched network. The Internet Engineering Task Force's Multi-protocol Label Switching (MPLS) group is working on this and likely will come up with a solution this year. Submissions from vendors such as Cisco, IBM and Toshiba, all of which are members of the MPLS group, could help restrict the use of virtual circuits to situations subscribers are willing to pay for, such as high-quality Internet access or VPNs. They also could create small, partitioned subnetworks where users could employ any protocols or address conventions they like, regardless of IP uniqueness or compatibility, without risk of collision with other users or security problems. "Could'' is the operative word here. The issue of VPN support isn't paramount with the MPLS group, and apparently it's not paramount with vendors in the switched IP space either. The first commercial product announced after Ipsilon's IP switching was launched was Ascend's IP Navigator. While Navigator supports partitioned VPNs and specific service quality to customers, users of the ATM switch products which Navigator runs on often are not aware of the features or how to use them. Other vendors with MPLS strategies, such as Cisco, have yet to even field a VPN-compatible product. Why not? Because of all the Ipsilon hype. We aggrandized the concept of IP switching so much that we forgot to tell people what was useful about it. That let Ipsilon raise capital and operate for two years when there was no chance the company would succeed. It also encouraged all of Ipsilon's competitors to avoid the knotty issues, even though the current MPLS approaches could resolve those issues in a wink. At least one competitor is likely to be winking furiously, and soon. Larry Lang, a key player in Ipsilon's heady flight and fiery fall, has returned to his former employer, Cisco. It's pretty unlikely that Cisco hired him back to continue his work on IP switching; more likely, Cisco is counting on Lang to apply what he learned from Ipsilon's failure to ensure that Cisco's tag switching succeeds. That's all it will take to open the MPLS floodgates, since the other vendors in the MPLS working group surely will have to follow Cisco's lead. Some may even try to anticipate a Cisco announcement to get a jump on the market. Either way, MPLS will rise to become the basis for building public IP networks in the 21st century, and IP switching can rest in peace.