Hacker News Reader

Comment by alphabeat | Hacker News Reader

I believe it was the OS from memory. This was around the time Accenture lead development wasn't working out a few years ago.

View on HN

I would be really surprised if it was the fault of the OS. I mean, Windows can be really fast, in the right hands, especially since they've introduced HTTP.SYS (parts of the HTTP server now run in kernel level). However, .NET being a garbage-collected, managed and JITted environemnt, it is maybe not the best tool for achieving these kind of response times reliably. Any serious and competent .NET developer would admit that.

In general, I would blame the incompetence of Accenture for the failure: bad decisions, bad architecture, bad management, using the wrong tools, crappy code.

Are they actually using HTTP for anything?

I have absolutely no idea.

> especially since they've introduced HTTP.SYS (parts of the HTTP server now run in kernel level).

Isn't it something Red Hat did a long time ago that was widely regarded as a Very Bad Idea?

And why would you use HTTP in this scenario anyway?

I don't know why it didn't work out for Red Hat but it works quite well on Windows. And honestly I don't know whether they use(d) HTTP or not, but given that HTTP is dead simple and as lightweight as a protocol could get, maybe it wouldn't be such a bad idea to use it for exchanging messages in this scenario.

.Net and newer windows have some really cool real-time features -- like deterministic garbage collection options, various per-thread priorities for scheduling, preemption and so on, making it a decent real-time platform. (for some definitions of real-time).

Reply on news.ycombinator.com