Hi Hassan,
Thank you very much for the quick response!
Can increasing the number of partitions really help in this case? All
partition data needed for calculations is already offloaded to disk, so
there are no partitions in memory. OneMessagePerVertexStore contains
vertexes for ALL partitions in
Hi Denis,
Thanks for trying out the new OOC design. As you mentioned, by using
message combiner we only keep one message for each vertex at all time. That
means the storage needed for messages is #ofVertices*(SizeOfVertexId +
SizeOfOneMessage). We studied several applications that use message
comb
Hello,
We are using OutOofCore functionality to perform computations over
huge graph (billions of nodes).
Recently we have faced a situation when all our workers stuck doing
nothing except performing System.gc() triggered from Giraph's
ThresholdBasedOracle. The intriguing point was that no memory