Mr. Clinton,
Interesting comments. I thought I might respond:
Mory has concerns about Cisco's ability to be price competitive more than
anything else
What Cisco has that should worry Mory is the huge installed base of routers that they have among enterprise customers. Especially at the high end where Cisco makes their $$, there is so much infrastructure built around Cisco enterprise routers and their specialized (and buggy) software that the costs of switching to non-Cisco are very prohibitive. I work for one of Cisco's largest customers, and they literally have us by the balls. We can buy Bay routers at 1/3 the price of an equivalent Cisco router, but it would be impossible for us to integrate them into our huge Cisco-based network.
This installed base of enterprise customers is Cisco's warchest. And as you suggest, Cisco can use this to wage a price war on Ascend's home turf in the carrier market. Since Ascend's enterprise business is weak/non-existent, Mory cannot counterattack in the enterprise market. Today, Cisco is truly invulnerable on their own home turf, the enterprise market.
This scenario is always a possibility if Ascend fails to stay one step ahead of Cisco in the carrier market. Ascend has been great at doing this, but how long can they keep it up? I don't know. The only permanent safe haven IMO is running into the arms of a protective Lucent.
Many carriers are still sceptic about the abilities of pure IP networks
to deliver a true QOS capability.
True, IP does not have true QoS, at least not yet. But ATM does not scale as easily as IP. QoS vs. Scalability. These are inherent tradeoffs.
Also, none of the carriers are using ATM to deliver QoS, at least not yet. Rather, it's used to oversubscribe bandwidth so that carriers can sell and resell the same physical circuit multiple times to different customers. This makes it cheaper for the customer, while the carrier better utilizes available physical bandwidth.
They are comfortable that ATM can deliver this
and if someone can have IP routing in this architecture with no bottlenecks
(alongside frame relay) then they've got all bases covered.
There is no question ATM can deliver this. The only question is can it be scalable to a carrier-sized network and larger. MPOA is one example of a non-scalable IP/ATM implementation. This is where Ascend's IP Navigator comes in. IP Navigator claims to automate and manage the complex and overwhelming process of mapping IP flows to ATM SVC's. So far, no one is using it yet.
When it comes to running
voice over ATM it's more bandwidth efficient to do this over native ATM than IP.
Some will disagree with you. There is RTP header compression which reduces the IP overhead for voice to about the same as ATM. You can fragment IP so that it looks just like and has the same characteristics as ATM.
The scenario where this is true is if you do VoIP and then put it over ATM, then you have the additional element of fragmentation which can cause up to 50% additional overhead. Other than that, VoIP can potentially be about as good as VoATM.
IP is ubiquitous, while ATM is not. It's much easier to do VoIP at a small branch office or home than to put an ATM concentrator at every location.
I believe both VoATM and VoIP have their roles. VoATM is better for trunking, while VoIP will be used at the edges because IP is ubiquitous. This translates into carriers primarily supporting VoATM in the backbone, while enterprise customers rely on VoIP. Once again, I give the edge to Ascend in the carrier network and Cisco for the enterprise.
bucky89 |
