GMS is a memory intensive application server, 300-400 MB is really not a lot here. Our recommendations for GMS are on the order of multiple GB. If you are looking for something lighter where you don't need all of the functionality a full deployment provides for a local deployment take a look into our
DataHub Lite Beta. There are quite a few different components involved with ingestion and if you are running it all locally then yes, you will see large memory spikes from GMS, Kafka, ElasticSearch, MySQL, etc.