While the comment might have come off as "ridiculous" I think the underlying theme is quite pertinent.
First, in response to Joel's comment, obviously Stack Overflow is personified as "Joel", the same way Apple is "Steve" and Windows is "Bill". What I find charming is your personification of it as collection of individuals who are are equally invested... it must be ponies and rainbows over there.
What's pertinent about the comment and probably at the heart of his "fascination" (mine too) is how this product is held up as a poster child for ASP.NET MVC, without giving the "whole story". The is very common of the Microsoft evangelists. Let's be clear, I am not accusing Joel of being a "loyalist" I think he is probably only pinned as one because of these evangelists running around trumpeting the use of ASP.NET MVC for this successful site. The true MS loyalists will never read this article or decide that Stack Overflow just "doesn't know about" the correct MS technology stack that could eliminate his need for any Open source.
So that's what's "fascinating". That an ASP.NET MVC application did not stay inside the reservation.
As for the right tool for the job nonsense, let's cut the bullshit. We all make religious decisions when we "start" a software project. You build out an idea your excited about in a language/platform your excited about. What's interesting here is that the real world of success has demanded an open source stack... and it could very well be cost and not capability. You are in FANTASY LAND if you think you would be paying a "couple grand" in licensing to scale out to what Stack Overflow must need to run its service. Add about 3 more zeros to that guess-timate.
End of the day, I don't think anyone regrets not going 100% open source you build it in what your excited about. The story I would be most interested in is what performance/cost analysis was done that steered them in the technology directions they took. I think we could all learn from that.
People rarely pay sticker price for this stuff. Right now SO is still using bizspark, but lets pretend we are paying sticker for our NY Datacenter:
* 10x Windows Server Standard R2 for the Web Tier (~1k per)
* 4x Enterprise OS (4k) for the DBs (were are going to give SO its own DB soon)
* 8 Sockets of SQL Server Enterprise R2 (27k) .
So 10x1000 +4x4000 + 8x27000 ~= 242k sticker price. Again people don't tend to pay sticker price for this stuff -- but this guesstimate is not $2 Million by any means.
As for the right tool for the job nonsense, let's cut the bullshit. We all make religious decisions when we "start" a software project.
I think you speak for yourself here. The fact that you can't even imagine that people don't make religious decisions at the start of a project is telling.
There are Java projects that use Oracle as the back end. Since Java is Free Software (technically at least) and Oracle is not, would you consider that to be a failure for Free Software?
Do you use a binary blob NVidia driver on your Linux computer? Would that be a case of not staying in the reservation?
You're a Rails developer, aren't you? twitter was hailed for a while as validation for Rails and Ruby in general. That is, until they realized it wasn't scaling and had to use something else. Failure for RoR? I guess they weren't religious enough (as you say) to stick with what they had and gosh darn it, make it work at all costs. After all, you can run your blog on RoR, just like you can on MVC. If you imply that using Redis on Linux to scale invalidates MVC as a viable platform, how is using TokyoCabinet (or whatever) to scale RoR any different? Unless of course your argument is purely ideological.
We make decisions on what technology to use based on what we know. On our areas of expertise and knowledge. If we pay X for something, it's because we believe that X is a good price for what we're getting back. If we decide to use Y because it's free, it's because it works the same as Z, which is expensive or limited.
If there are "true" MS loyalists out there who make decisions based on their adoration of the company, then that's their problem. More power to them and all that.
And by the way, if you feel that this represents a form of "capitulation" for the MS stack, then you need to go talk to MS about it, not StackOverflow. Because they made their technical decisions based on practicality, not religion (as you say). And besides, if it's supposed to be so embarrassing, why mention it at all? No one is forcing them to disclose what platform they happen to be running their key value store on.
>You are in FANTASY LAND if you think you would be paying a "couple grand" in licensing to scale out to what Stack Overflow must need to run its service. Add about 3 more zeros to that guess-timate.
Couple grand = $2000
Add 3 more zeros:
$2,000,000
Stackoverflow is paying $2 million in software licensing fees! Wow, using the Microsoft stack is expensive!!
First, in response to Joel's comment, obviously Stack Overflow is personified as "Joel", the same way Apple is "Steve" and Windows is "Bill". What I find charming is your personification of it as collection of individuals who are are equally invested... it must be ponies and rainbows over there.
What's pertinent about the comment and probably at the heart of his "fascination" (mine too) is how this product is held up as a poster child for ASP.NET MVC, without giving the "whole story". The is very common of the Microsoft evangelists. Let's be clear, I am not accusing Joel of being a "loyalist" I think he is probably only pinned as one because of these evangelists running around trumpeting the use of ASP.NET MVC for this successful site. The true MS loyalists will never read this article or decide that Stack Overflow just "doesn't know about" the correct MS technology stack that could eliminate his need for any Open source.
So that's what's "fascinating". That an ASP.NET MVC application did not stay inside the reservation.
As for the right tool for the job nonsense, let's cut the bullshit. We all make religious decisions when we "start" a software project. You build out an idea your excited about in a language/platform your excited about. What's interesting here is that the real world of success has demanded an open source stack... and it could very well be cost and not capability. You are in FANTASY LAND if you think you would be paying a "couple grand" in licensing to scale out to what Stack Overflow must need to run its service. Add about 3 more zeros to that guess-timate.
End of the day, I don't think anyone regrets not going 100% open source you build it in what your excited about. The story I would be most interested in is what performance/cost analysis was done that steered them in the technology directions they took. I think we could all learn from that.