John - 3 things need to happen.
1- "platform" software needs to exist and be optimized for Itanium. That includes the base OS, key database and middleware components, and the like.
2- Second tier applications - those that build on the base platform - need to exist
3- some big players - customers - need to dive into the pool.
A few years back, Intel themselves said that they thought the cycle was 18 to 24 months. They have since spent a lot of time, effort and money to get all three of the goals accomplished, with only modest success. The hidden issues with Merced - mainly the fact that almost nothing except the instruction set, and not all of that, goes forward to McKinley - will keep Merced from being anything but a prototype development platform. Everyone in the game has known this for years. So the new play is to get as many of the Merced machines into the second and third tier as possible, so that development tools, applications components and so on will exist when McKinley rolls out.
The way things are going, I would say the 24 month clock starts about now - in other words, we will see gradual increase in interest in Itanium, but there will be little volume until the McKinley machines are out for a while and the proof points come back. So my guess is summer 2003... |
