Re: Why ASP.NET gets no Respect

[Nathan Andelin <nandelin@xxxxxxxxx>]

> From: "Dean, Robert"
> I was thinking the same thing about that study.  Some of the tuning
> decisions seemed arbitrary, and others seemed to be designed to
> deliberately hamstring the WebSphere-on-POWER system.

Possibly.  It seemed odd that they configured only 2 instances of Websphere on an 8-way Model 570 and set the maximum thread count at only 50 - while they configured 4 instances of Websphere & ASP.Net on a 16-way Wintel blade, for example.

> Who in their right mind would actually build WebSphere on AIX and then
> force it to talk to DB2 on Windows instead of DB2 on an AIX partition?

Right.  And if price/performance were the objective, you'd probably run both in the same partition of the Power Server.

But Microsoft came up with a fantasy configuration of front-ending two 16-way database servers with an 8-way Power Server and a 16-way Wintel Blade.  Then only reported (and actually highlighted) the price/performance of the application servers - a $260K Power Server vs. an 87K Wintel blade.

A thoughtful analyst might consider that the real expense is in database servers - not application servers.  But it would not be in Microsoft's interest to compare the price/performance of a single Power Server running a mixed workload to three Wintel Blade servers running distributed workloads - and showing database server costs.

No, Microsoft would rather highlight a specific price/performance metric that sends a manipulative message to possibly clueless executives.

Forget other TCO factors such as the extra cost and performance drain applicable to virus protection on Wintel servers - just to mention one.

Nathan.