Hi All,
Does anyone know a good resource (website, book, etc) on things to think about when scaling a website? For instance, I am looking for materials on:<p>1) How a scalable website is organized in terms of infrastructure (having a database server, having a web server, etc)?<p>2) Best practices for designing the database for a large website to prevent single-points-of-failure<p>3) What web services out there allow for easy scaling. I have heard of slicehost, Mosso, EC2, Google App Engine (any others? We use PHP)? Do you have to configure your own load balancer etc for these services?<p>4) Other useful things to think about when designing a scalable web application.<p>Thanks all
<a href="http://highscalability.com/" rel="nofollow">http://highscalability.com/</a> has some fantastic articles.<p>The best tool to learn how to use is memcached - it has clients in most major languages (And even some obscure ones) and when used properly can remove a lot of the typical problems for you. Part of what I found with memcached is that it forces you to think in terms of encapsulated data. Instead of thinking about your database all the time, the database becomes a tool. What's more important however is consumable bite sized chunks of data that your application is using, and finding ways to get to them quickly.<p>Stay away from apache: it's far to slow and unwieldy for most scaling. I'm a huge fan of LigHTTPD (<a href="http://lighttpd.net" rel="nofollow">http://lighttpd.net</a>) but a lot of people also swear by Nginx (<a href="http://nginx.net" rel="nofollow">http://nginx.net</a>).<p>Depending on your language... find ways to tune away any of the runtime slowdowns such as translation to byte code. With PHP find a byte code cacher, learn it tune and and rely upon it. The LigHTTPD guys make a great cacher called XCache (<a href="http://xcache.lighttpd.net" rel="nofollow">http://xcache.lighttpd.net</a>) which I've used on a few big sites. It plays nice with Lighttpd and is actively maintained.