I have a dataset where each row has three features . The row represents a single edge (from source to destination) and the ID of the trace it belongs to. Note that these traces are invocation of microservices collected from an observability tool such as Jaeger. So there could be multiple traces (with different traceids) but the same edge connections. I want to achieve the following: 1.Parse each trace separately into a graph. 2.Group graphs which are the same structure. 3.Dump a representative graph from each group and the count that graph occurs in my dataset. Note that I have 2 million such graphs (average number of nodes in each graph is 15). Is GraphX suitable for such a problem?

I am currently parsing this as an edge RDD but I am not sure how to parse each graph separately. Should I have multiple graph objects for each graph?