parallelism
The optional parallelism
parameter can be set to reduce the number of threads used by the bulk load process.
Allowed values:
LOW
– The number of threads used is the number of available vCPUs divided by 8.MEDIUM
– The number of threads used is the number of available vCPUs divided by 2.HIGH
– The number of threads used is the same as the number of available vCPUs.OVERSUBSCRIBE
– The number of threads used is the number of available vCPUs multiplied by 2. If this value is used, the bulk loader takes up all available resources.This does not mean, however, that theOVERSUBSCRIBE
setting results in 100% CPU utilization. Because the load operation is I/O bound, the highest CPU utilization to expect is in the 60% to 70% range.
Default value: HIGH
The parallelism
setting can sometimes result in a deadlock between threads when loading openCypher data. When this happens, Neptune returns the LOAD_DATA_DEADLOCK
error. You can generally fix the issue by setting parallelism
to a lower setting and retrying the load command.