I liked a recent episode where Richard Campbell and Scott shared war stories about scaling large websites.
Excerpts:
I'm consistently surprised at how many sites don't have a geographic failover.
So, what is the difference between a performance problem and a scale problem?
Well, I think most people see scaling problems manifest as perceived performance problems.
When someone tries to start aggressively caching, whether that be output cache or formally caching from a business perspective, there's this Cartesian product that happens.
I like the idea of being able to attribute something or mark something and have it handled because this is a common pattern and anytime something is a common pattern, I expect the best practice to be embodied within the library.
..if you start thinking about performance early on, you always get accused of being a premature optimizer.
dasBlog is an interesting application because we don't have a database.....Everything in dasBlog is stored in XML.
You know, what I like to do is I turn it (ViewState) off completely and then turn it on.
While, Microsoft, now that I'm starting to understand what they think about, they have to make that distinction between speed of development versus the expectation of the developer. They wanted ASP.NET to just be one of those things where I can drag a button onto a page and it works.
One of my favorite defaults and I've talked about this in sessions all the time that people a) don't know about, but b) when you really think about it realize it's a good default, but seems totally non-sensical is ASP.NET's default worker process restart is 23 hours. That's the default setting for a cold install of IIS. Every 23 hours, restart the worker process.
The MSDN "Geek Speak" Webcast series is another good resource.