72 minutes is the interval between nodes taking ledger snapshots… but that normally causes the load & CPU usage to spike, not the memory usage.
I don’t know what “Memory Utilization” is in this graph, but it’s certainly not your percentage of 16GB RAM being used, since the Cardano node needs at least 6.5GB of memory space to perform normally at all. But it does sound like your system is running out of memory if the node process is getting killed. You could try:
1 - upgrading from cardano-node 1.34 to 1.34.1 since as I remember 1.34 was a tag used internally by IOG & not an official release… maybe there are some memory bugs that come out when doing the ledger snapshot?
2 - you can change the variables for the ledger snapshot interval if you need to do any more testing: