What can be optimized for Flume in the process of data collection

• Flume's double-layer collection (first-tier data collection, second-tier data aggregation)
• Channel can use Kafka channel, which has the advantages of File Channel and Memory Channel; it can also have both offline and real-time computing.
• Flume is replaced by a lightweight log collection component logstash (L in the ELK suite), or a lighter set of filebeat
• Flume interceptor, you can take extra data into consideration and only keep the Json string