Hi all. We currently design a topology that has custom processors, some with sufficient processing, for example, an ASN1 decoder. We only have one server with 56 cores and 252 Gb of RAM. Initially, this topology ingests data through NIFI to then perform the processing in STORM and then generate the output again using NIFI. The problem is that STORM processing has memory errors and the latency is quite high. What values can be configured to solve this problem, thanks. Here are some screenshots of how the solution is implemented.
Thanks for your comments