ASP.NET is moving at an incredible pace, the new vNext bits are planned to be officially cross-platform and you can support any number of database backends from a .NET codebase.
I also challenge your assertion that better people seem to use other platforms. You can find people marketing themselves as one thing but preferring another in any stack. This is not indicative of anything more than someone needing a job and meeting the requirements of a posting.
Licensing costs can vary. Between BizSpark, DreamSpark and other programs you can get dev tools out of the way quickly. They also provide Windows licensing for the period you're in the program.
The areas where you can get in real trouble as a startup with Microsoft are the add-on product stacks like SQL Server and BizTalk (shudder). I would avoid these in a new product but do not fear building with the .NET platform.
Asp.net v.next is currently a bag of promises. I wouldn't put a product near it for a long time. I back this assertion up with the promises of Velocity, EF4, SilverLight and WF4, all of which were disasterous piles of immature crud that disappeared after a bit leading to massive rewrites. Even MVC has a patchy history and numerous problems in it now (crazy API churn, attribute lifecycle/scope, memory ceiling, routing performance etc are all ones that I've spent days on...). Generally the entire web and enterprise teams have knocked out low quality rubbish for years.
People who advertise themselves as .net developers (as I do) know how to get paid a lot, not necessarily deliver the best solution for the money. There are those of us however who have fingers in many pots who know how greener the grass is and how organisations would benefit from a change and are taken on to fix the tech stack.
Agree with your assertion about BizTalk but mno further.
Can you go into this a bit? I just upgraded a MVC2 site to MVC5 and didn't have to change much.
I know the DI bits changed from MVC2-3 but all the existing extensibility methods worked.
Oracle? SAP?
Oracle tooling I.e. JDK, Netbeans, Oracle Linux (repackaged RHEL) etc is 100% free. VS isn't.
How can this happen if you already have the software licensed? Shouldn't the result be that you owe them $0?
Action pack is reasonable value but you can't use it to deploy production kit on. Not only that the definition of internal use is so vague it's a risk.
We got an audit. It wasn't cheap to sort out what we thought was in compliance.
At that point we stopped playing word games and threw our devops infrastructure on Linux.
Internal use licenses are for use in a production environment for general internal
business purposes and not for any type of commercial purpose. For example, installing
the Windows Server operating system and Microsoft Exchange Server to set up an email
system that you can use to send business-related email is acceptable. However,
production use for external commercial purposes, such as hosting a commercial website,
is outside the scope of these rights.
Customer demonstration licenses can be used for demonstration purposes only by partners.
Sales and marketing employees of the organization can use this software to showcase
products to your customers, but demonstration products cannot be installed on customer
hardware or infrastructure and must be used with partner supervision.
Internal-training licenses can be used for training internal employees only. These
licenses cannot be used for customer training or for any commercial purpose.
MSDN subscriptions allow designated employees of your organization to use the available
software for the design, development, testing, and demonstration phases of application
development.Straight away we have ambiguity.
Can you run your staging site on it for demonstration purposes?
More ambiguity.
All it takes is a small mis-step.
Can you install JIRA on that copy of windows 2008 server R2 and access it via VPN?
Depends on what version of JIRA you're talking about. As long as it's supporting your product, be it purely internal (project management) or customer related (issue tracking and CRM) I'm pretty sure it falls under the first paragraph. Note that it states internal business purposes, not on site use only. CRM is not "commercial purposes" in it self, and resembles email example. Can you run your staging site on it for demonstration purposes?
It clearly falls under the second paragraph.
It won't make a blind bit of difference initially but if you expand, so will tooling and deployment costs and investors don't like that. Anything that takes off the bottom line is a problem. Microsoft licensing, particularly dev tools and SQL is incredibly expensive and an order of magnitude more expensive on deployment and diagnostic time than anything else.
We're like that now. Our SQL licensing is shaping our architecture not on technical merit but avoiding core license costs. That is beyond bad but we can't justify a £200k spend on licenses and kit to get rid of 2008R2.
Inevitably bits of Python are now appearing around the edges of the product and trying to stab the core. There are a couple of postgresql machines doing ancillary work. All the staff are slowly moving their operational task over because there isn't a purchase process.
This is the price of success though :)
Edit: just to add that the last two companies I worked for are Python/java/postgres houses now after being end to end MS outfits. The staff change over the lifecycle of a product. You can afford better people after time and better people seem to prefer to use other platforms even if they market themselves as .Net people.
Edit 2: also once you're locked in, volume licensing looks cost efficient. Then after a couple of yeara you get a call from the sales guys at MS who want to do a 'soft audit' (which it says they can so in the VL agreement). Then battle commences between MS, yourselves and the disty who handles your VL over how much you owe. This usually starts at £150k-ish for an SME and your ops team will be down for days whilst they work that down. Our liability was less than £10k which is within the VL safe zone but other companies are less lucky. Also the VL settlement says you can't discuss it but fuck them - the company this was against is dissolved.
Actually I look at my post and I couldn't possibly recommend it. If we'd started differently we could afford to be employ and sponsor postgres core team staff for example and gave second to none support and a positive impact on the community. Instead, pockets are lined elsewhere.