Currently each Jira node is involved in synchronisation DVCS data with git.
When incomming HTTP traffic (webhooks) we can steer with load balancing, than we are no control over outgoing traffic.
As a outgoing traffic we understand hourly synchronisation and messages consumption as well.

There were some customer questions if they can turn off DVCS synchronisation on specific nodes to provide stability for UI users, when dedicated nodes would handle whole synchronisation process and HTTP requesting.