If the occurrence time is relatively fixed, then continuously show full processlist to see which suspicious statements may cause high traffic, full table query without filter conditions, etc. If the occurrence time is not fixed, then monitoring is required: 1. Database monitoring, use monitoring tools to automatically perform the show full processlist operation and record it in a file. doDBA tool can achieve this. 2. Capture traffic packets, solve problems head-on, do packet capture monitoring, and analyze what statements cause traffic after packet capture. Of course, here are just general directions, you still have to study the specific operation details yourself
If the amount of written data is high, you need to check whether there is a scheduled job to import data.
At the database level, you can consider using tools to count the execution frequency of insert statements. If the execution frequency is too high, it may lead to high write IO.
If the occurrence time is relatively fixed, then continuously show full processlist to see which suspicious statements may cause high traffic, full table query without filter conditions, etc.
If the occurrence time is not fixed, then monitoring is required:
1. Database monitoring, use monitoring tools to automatically perform the show full processlist operation and record it in a file. doDBA tool can achieve this.
2. Capture traffic packets, solve problems head-on, do packet capture monitoring, and analyze what statements cause traffic after packet capture.
Of course, here are just general directions, you still have to study the specific operation details yourself
If the amount of written data is high, you need to check whether there is a scheduled job to import data.
At the database level, you can consider using tools to count the execution frequency of insert statements. If the execution frequency is too high, it may lead to high write IO.