Hi. I am having data of around 4B Vertices and 13548192791 edges, say with avg outgoing edge of 4, so total 4B*40 bytes= 160G , When I load this into memory across 165 mappers each with 6G mem, all memory get occupied, I am confused here as we have around 1TB space in memory. What is its thats taking ~800G , is it the mapping information stored ? is there anyway of doing profiling over it ?
Thanks -- Best Regards, Jyotirmoy Sundi Data Engineer, Admobius San Francisco, CA 94158