Server-to-Server vs Download-and-Reupload: Why the Difference Decides Your Transfer
· 4 min read · Updated
There are two ways to move files from one cloud to another. The common way downloads every file to your computer and then uploads it to the destination, which leaves a full copy on your disk. The other way, often called server-to-server or a direct transfer, streams the data from one service to the other without staging a full copy on your disk, so local disk space stops being the limit.
With ten files you won't notice the difference. With ten thousand, it decides whether a migration finishes or stalls halfway through a terabyte.
The label is looser than it sounds. A literal server-to-server copy, where the two providers hand the bytes straight to each other and nothing passes through you, only happens in narrow cases: within a single provider, or between two services that specifically support it.
Blober uses a server-side copy where one is available. For a copy or move within the same account (the same bucket or storage account, and the same region for S3), it first has the provider copy the files, so no data passes through your computer. Seafile and Drime Cloud work this way inside one account too. Between different accounts or providers, or if the server-side copy fails, Blober makes a streamed copy: the data flows through the app without being staged on your disk, so disk space stops being the limit. In a streamed copy the bytes still travel down to your computer and back up over your own connection; what you save is the full copy on your disk.
Drag a cloud folder to your desktop and back into another service, or let a sync client mirror it locally, and you are doing a download-and-reupload. It works, and for small jobs it is fine.
At scale it breaks in three ways. First, disk: a 2 TB move needs 2 TB of free space, which most laptops do not have. Second, time: every byte makes two trips through your machine, often one file at a time. Third, fragility: a sync client working through thousands of files will choke, create duplicates or stall, and you end up watching over it.
A direct transfer connects to both services and streams data from the source to the destination without writing a full local copy to disk. Because the files are not staged there, you can move far more than your free disk space.
The bytes make the same two trips, but many files move at once instead of in one slow queue, and that is what gets a big library finished. If a transfer drops partway, resuming it continues with only the files that did not make it, so the whole job does not start over.
In a streamed transfer, the data still travels over the connection of the computer the tool runs on, so your bandwidth sets the pace. Egress, what a source provider charges to read your data out, applies whichever tool you use, so plan the move as one pass instead of repeated pulls.
A separate question from disk space is who handles your data. Some cloud-to-cloud services route your files through their own servers, which means a third party touches them. Blober is a desktop app that runs on your machine with your credentials, so your files are not parked on someone else's server between the two clouds.
What is a server-to-server transfer? Strictly, the providers move the data themselves so it never touches your machine, which only works in specific cases, such as a copy within one account. Blober uses that server-side copy within the same account. Between different accounts or providers it streams the data through your own machine, as most "server-to-server" tools do through theirs: the files are never staged as a full copy on your disk, though they travel over your connection.
Do I have to download files to move them between clouds? Not to your disk. Between providers, Blober streams from the source to the destination without writing a full copy to your disk, so disk space is not the limit.
Why is my cloud-to-cloud transfer so slow? A download-and-reupload often runs one file at a time. A parallel streamed transfer still sends every byte through your machine, but it keeps many files moving at once, which is far faster for large libraries.
Does a direct transfer avoid egress fees? No. The source provider charges egress for reading your data out, whichever tool you use. A direct transfer saves the full copy on your disk, and the egress bill stays the same.