Social networks often require the ability to perform low latency graph computations in the user request path. For example, at LinkedIn, we show the graph distance and common connections whenever we show a profile on the site. To do this, we have developed a distributed and partitioned graph system that scales to hundreds of millions of members and their...
