How are snapshots and transaction logs used in replication for ensuring no impact on the workload?
As an overview, snapshots and transaction logs are logs and records for all queries and transactions run against the cluster.
When utilizing replication in any capacity, instead of querying the database (increasing the size of the cluster's workload) to collect data, the replica will download the snapshots and transaction logs from the "primary" database and use those files to replay and replicate the data.
This process minimizes performance issues for the cluster due to replication processes that would query and consume throughput for the database.
The Snapshot Database Command will force creation of a new snapshot before it would otherwise naturally be triggered.