You should allow enough time for one version job to be captured before beginning another epoch for that job. If a versioning process cannot finish before the next process is scheduled to begin, the archive server continues to version the job and skips other scheduled processes until the job finishes.
For example, suppose that you configure a versioning job with a one-minute epoch for a source volume that contains an amount of data that takes 5.2 minutes to archive. The version process begins at time 0. At each 1-minute interval thereafter, the next scheduled versioning checks to see if the job is already running. If it is, the job does not begin. In this example, the scheduled versioning processes do not occur at 1, 2, 3, 4, and 5 minutes. The next versioning process begins at 6 minutes. Although the user expects to have file versions available at 1-minute intervals, the file versions are at 6-minute intervals.
To avoid such problems, consider the following configuration alternatives:
Set epochs realistically, based on the amount of data to be versioned from the source volume and the bandwidth of the connection between the source server and the archive server.
Filter the data on the source volume to exclude unnecessary data from being versioned.
Divide data on the source volume to create multiple volumes. Then you can configure multiple jobs to run concurrently on the same or different archive server.