Simple is Beautiful

It is safe to assume that most organizations that deal with scale, large amounts of data, jobs parallelization, all things distributed computing in general, are replicating what was made available by Google ( by means of papers and tidbits published on various site ). Which is a good thing. Google's research has combined common sense, pragmatic insight and novel ideas into beautiful systems. Yahoo! is effectively replicating the whole 'Google Distributed Computing Stack', by investing heavily on Hadoop ( consisting of a distributed FS not unlike GoogleFS and a MapReduce implementation ). In addition to that, they are building a system similar to Google Chubby (distributing lock service based on the Paxos algorithm ), a centralized jobs manager and so on. Its a great thing they are not trying to conceal the fact they are following Google's lead for the sake of me-first and rivalry.

As part of my research for this small project I have been working on, I got into studying Oracle Database's internals. I used to dislike Oracle with a passion. Bloated, over-complicated and over-priced. It turns out Oracle is really beautiful architected. From the way data-blocks are handled, up to transactions locking and consistent read (CR) facilities, its all so well thought out and so nicely put together as whole. There is also a ton of information available on the way everything works, that one could "easily" build a similar product just by implementing what is clearly documented.