It is complex, but it's easy to game the costs to make Microsoft look more expensive than it actually is.
If you look at list pricing or in the US (I imagine Germany has something similar) government pricing, you can spec everything with the highest cost "Select" pricing, (which is fixed cost) or other schemes which will be higher.
If you compute the cost over 5 years and want Microsoft to look bad, you spec things with non-transferrable OEM pricing for servers and other products. You buy Windows server in the "Standard" edition and use VMware.
If you want Microsoft to look "good", you spec the pricing based on 3 year enterprise agreements with platform discounts. (Platform discounts are incentives to license Windows, Core CAL and Office together). With an enterprise agreement, you amortize the license cost over 3 years, and pay only maintenance year 4 and 5 (typically 20-25% of the initial annual cost), so you can make the Microsoft TCO much better. On the server side, you buy Windows Datacenter and System Center, and use Hyper-V.
Both approaches are truthful, but will yield vastly different results and will be accurate if you follow through. The tough part about Linux/OSS vs Microsoft, especially for user-facing stuff, is that many of the costs are people costs, and can be harder to predict.
If you look at list pricing or in the US (I imagine Germany has something similar) government pricing, you can spec everything with the highest cost "Select" pricing, (which is fixed cost) or other schemes which will be higher.
If you compute the cost over 5 years and want Microsoft to look bad, you spec things with non-transferrable OEM pricing for servers and other products. You buy Windows server in the "Standard" edition and use VMware.
If you want Microsoft to look "good", you spec the pricing based on 3 year enterprise agreements with platform discounts. (Platform discounts are incentives to license Windows, Core CAL and Office together). With an enterprise agreement, you amortize the license cost over 3 years, and pay only maintenance year 4 and 5 (typically 20-25% of the initial annual cost), so you can make the Microsoft TCO much better. On the server side, you buy Windows Datacenter and System Center, and use Hyper-V.
Both approaches are truthful, but will yield vastly different results and will be accurate if you follow through. The tough part about Linux/OSS vs Microsoft, especially for user-facing stuff, is that many of the costs are people costs, and can be harder to predict.