This leaves you in a spot. Let me try to throw some light on this issue. You most certainly can migrate many applications from Windows NT to Linux, but that is only a small part of the total picture.
What makes you belive that in an eyeball-to-eyeball comparison your boss will in fact choose to adopt SuSE Linux, even if Linux will significantly out-perform Windows NT? This is not a trivial question. The key questions that will drive your company's choice of platform are very complex. Companies choose platforms based on complex criteria that includes:
- Personal preferences
- Perceived support arrangements
- Fear of failure
- Software availability
- Prestige claims
- Follow the "sure-thing" syndrome
To be clear, it is not difficult to produce benchmarks that will favor either Linux or Windows Server 2003. The only valid performance comparisons are those that accurately mirror your particular applications being used under site-specific load cycles. You are not likely to find that from generic comparative benchmarks.
You need to find particular comparative data that will be of interest to your management. Does you company have multi-vendor supply policies? How does that play for a single vendor for a MS Windows solution? To what extent is your company possibly exposed to monopolistic practices? How sensitive is your management to risks posed by viruses and worms? What economic values can you quantify to build an overall case for the IT platforms your company should use?
My advice is to forget the apples-to-apples comparison and instead focus on the overall value proposition. If that evalation is overall sustainable, then perform a side-by-side performance evaluation -- if it is still needed. You may find that this becomes a non-issue when all the intangible, political and emotional factors have been taken into account.
This is a well respected saying: "People buy by emotion and then justify
that decision using logic." If you attempt to reverse the order, you could end up wasting a lot of time and never gain management buy-in.
This was first published in March 2004