Kurt Shintaku's Blog

So I talked to my wife about this in great detail.  She’s a former Intel veteran that worked on aspects of the Itanium 2.  She soundly convinced me that this actually makes complete and total sense and doesn’t necessarily minimize Itanium’s importance or intent at all.

 

Why?  Well, Itanium was designed to beat Sun at their own game:  RISC-like processing.  It was designed to scale up and process instructions ‘better than the Sun UltraSPARC processor… better… stronger… faster!’.  It would process instructions similarly to RISC processors lending its abilities to handle big Line-of-Business UNIX job-based applications well, making UNIX–>Windows (or more appropriately, UltraSPARC–>Itanium) migrations architecturally simpler than moving to x86/32-bit processors of today.

 

And this change in support doesn’t change this agenda at all.  Nobody was going to use Itanium today to make File & Print servers of domain controllers or Windows Media Servers… so why spend all the money and time and labor for Microsoft to test for those scenarios?  Indeed, if Itanium is for big workhorse jobs that are currently running on UltraSPARC UNIX-based systems, then this support policy doens’t change anything.  Itanium is still a niche product, and will continue to be so until such time as 64-bit applications are the norm, x64 runs out of gas scalabilitywise, and Itanium technology takes over where x64 leaves off.

 

If it gets too expensive to support Intel, will drop it, and it’ll be NO BIG DEAL.  Why?  Because the investment in Itanium is ‘sunk costs’.  It can’t be recouped and the company is, let’s face it, doing just fine beating the snot out of AMD.  In fact, keeping Itanium alive with it’s own fab facilities might be MORE costly for Intel considering the cost of maintaining said facilities for such a low volume item.  Remember:  This isn’t Sun we’re talking about… Intel and Microsoft live and die by the mantra, "Low prices, high volumes".

 

Intel will be just fine.

 

—————–

Wow.  Big news hit today.

 

Windows Server 2007/Longhorn Server will be squarely focused at big workload services like SQL Server, other scalable databases, and line of business applications.  Traditional services will not be supported. This includes fax server, Windows Media Server, Windows Sharepoint Services, file/print server, and Active Directory domain controllers.

 

I’m still trying to wrap my head around this.  I always assumed that as time progressed, Itanium would be the successor to the x64 Extended architecture meaning that long term Itanium would be the goal and x64 would be a bridging technology.  I didn’t think anyone would retract support for Itanium for specific workloads.

 

I’m sure this is to best use our developers and our resources being that Itanium is such a small market and testing all these workloads for every revision and service pack would have been a lot of effort for very little actual customer usage but WOW.  It looks like x64 is here to stay and Itanium might be marginalized.

 

That being said, work is still being done to fully support Itanium Windows Server Longhorn in the way of management and other requirements.  Windows Server “Longhorn” for Itanium-based systems will continue to support client-side functionality for administration, management and server utilities so that’s good.