Replications randomly failing with the following, or similar errors:
IOException: Unable to read data from the transport connection: An existing connection was forcibly closed by the remote host.
There are a number of possible causes for such errors.
Generally, adjusting timeout settings on both source and target cores as per the following KB resolve them.
However, there may be underlying bandwidth constraints which, even with these changes, do not help to resolve the issue.
Another possible cause of these errors however, is a known issue with Broadcom NetExtreme 1-gigabit network adaptors.
Please see the following MS article in relation to this issue: