actually its abit confusing
. Ts wanted to making a "facebook"
but its small. He wanted to make a facebook website when facebook is a network/system...If its small and dont employ technology used for commercialized business(data mining,complex algorithm),then it shouldn't be link at all to facebook. Look more like a forum for a community to me.
By being unclear i just took it that he wanted to expand something to probably facebook size eventually(if not why state facebook as an example and choose to make a social website). And if thats the case its probably better to plan big first with distributed design etc.
Well nowadays when people talk about social sites, the first thing comes to mind naturally is Facebook. So I'm not surprised it's used as an example. But before Facebook, there are quite a number of other social networking sites such as Hi5, Friendster, MySpace,Tagged, Multiply, etc. I wouldn't call them small scale, but likely not as large the scale as Facebook today.
If we just look at the core requirement of a social website, it's not all that different from any typical web application. That's all I'm discussing. Even if it grow to a medium scale, it still can function pretty well using scalable database and multiple web servers and application servers without the involvement of queue systems, distributed data mining solution etc.
True enough, if we look at how a forum functions, it does shows a good amount of social interaction, after all it is designed as such. The fundamental differences I suppose is the main design.
Forum are hierarchical in nature. It has the concepts of rooms and sub-rooms. Users have profiles but they merely leave messages in the forum, or at times private messages dedicated to some members.
The difference in Facebook kind of social interaction is a directed graph. It's not hierarchical. You may form cluster given a set of users(nodes), but ultimately it's a large flat graph-like design where edges are basically relationships between users. You can represent a graph-like data structure inside a database schema, that's no issue.
Even users' resources such as photos, videos, applications are part of the graph. They have relationship to the nodes, but also have edges out to other nodes in permissions etc.
Of course, that's simplistic, because it will be a large effort to traverse the graph to query, but that's really another topic. If it's small scale, it's alright. There are ways of optimisation such as using caching, redundant data representation to make the traversing faster.
All I wanted to say is to grow to the kind of scale of Facebook definitely requires a lot of planning along the way, introducing new technologies, optimizing design and infrastructure. It's not like Facebook becomes a giant overnight. I'm quite sure employees of Facebook improved over time, making mistakes here and there.
