|
The impact on the DB should be similar, if not identical, to syncing a new secondary.
The backup agent scans every Collection in every Collection (minus any blacklisted namespaces that they configured). Data is read in $natural order, so I/O should be sequential. From an impact perspective I expect I/O reads to spike, data to get evicted from the working set, and outgoing bandwidth to spike.
There is no throttling in the agent, it consumes data from the DB as fast as it can deserialize, compress, and push data to BRS.
|