A good analysis – but can I it a bit further?

As human beings, the natural temptation is to paint ourselves into corners. Back in the 90s the establishment in carriers (and some vendors) was far too slow in seeing the potential of IP, partly because it challenge existing thinking and partly because there was an ‘IP community’ that was generally younger, dismissive of old ways of doing things and didn’t wear ties. This was the time of the ‘netheads’ vs ‘bellheads’ debate.

Luckily some people (of course I would say including you and me) didn’t get stuck in this sterile debate and were able to (hopefull), exploit the opportunities that the new approaches brought. One thing that seems to have been lost however is that the previous generation were used to the implementation of new network working technologies; from PDH to SDH to DWDM, from X.25 to ATM to FR to IP to MPLS.

These days there seems to be such an IP ‘hegemony’ that people appear to be less open than before – you yourself use the word ‘heresy’ in your post above. I am concerned that removing funcationality from the IP cloud is seen in this way by many people whose careers have been based on an in-depth understanding of IP from their undergraduate days onwards.

Maybe I am overstating the case, but it feels like it much harder to get the ‘best’ thing done these days and the simplicity that IP was supposed to provide has been over-engineered away. A lot of people do seem to have painted themselves into a corner on this issue, sometimes without even realizing it. Any ideas on how this locked-down thinking can be challenged / changed without getting into an us & them situation again?

I remember having the debate with COLT’s head of engineering (never shy about stating his opinions!) about the deployment of TE on the MPLS network. “We own the fibre, so why bother? We’ll just throw another wavelength onto the network.”

Similarly, on Metro Ethernet services, we looked at deploying C**co’s RPR technology (Resilient Packet Rings) as an alternative to the far simpler, far cheaper, and already existing Ethernet over SDH services offered by the likes of Nortel. As usual, Cisco had done a fine sales job on the senior management team (who understood nothing about the undelying issues – neither technical nor commercial) and had wound up half the product management team looking for customers to buy an expensive RPR solution. When you look at the business case for deployment of these technologies, it just doesn’t stack up … more cost for the network kit, more complex to manage and less scalable. Plus you would need to spend ages explaining to customers how they would get guaranteed bandwidth out of the service, and deal with security concerns.

Plain old Ethernet over SDH is far simpler … customers would instantly “get” the concept of nailed-up point-to-point SDH slots. If it ain’t broke, why fix it?