Replication process

[Joseph Anderson]

Hello,

I ran into something yesterday and want to confirm something about the replication process. Is it correct that during the normal aip storage process, the aip is created and uploaded, but then if there replicator set up, that aip is then downloaded back to create the replica? Is this correct?

The reason I ask is because this is probably fine and good if you are using the local filesystem for storage, but if you are using external storage (say aws s3), you can incur data transfer out fees to redownload the aip that you've just uploaded. Also it doesn't seem generally efficient because the original aip is already on the storage service local filesystem before it is uploaded. Why not just use that local copy to make the replica?

Thanks,

Joseph Anderson

Thanks,

Joseph