scp vs rsync: Choose the Right Tool for Fast, Efficient File Transfers

Core principle differences

scp – simple copy over SSH

scp

uses SSH to copy files. It always transfers the whole file regardless of whether the destination already has the file or not.

How it works :

Establish an encrypted SSH session.

Read the entire source file and stream it to the remote host.

The remote side writes the data, overwriting or creating the target file.

Advantages :

Syntax identical to cp; minimal learning curve.

No extra packages required on most Unix/Linux systems.

Works wherever an SSH daemon is reachable.

Limitations :

Transfers the full file each time – inefficient for large or frequently changing data.

No built‑in incremental or resume capability.

rsync – incremental synchronization

rsync

synchronizes files by sending only the differences.

Algorithm (checksum‑transfer) :

Remote side computes a checksum list (size, modification time, hash) for its files.

Local side receives the list, compares with its own files, and determines changed blocks.

Only changed blocks are transmitted.

Remote side reconstructs the full file from the received blocks.

Advantages :

Significant bandwidth and time savings for large or repeatedly updated files.

Supports exclusion patterns, attribute preservation, compression, and resume.

Can operate locally or over SSH for remote hosts.

Limitations :

Richer option set; syntax is more complex.

May need to be installed on minimal systems (e.g., yum install rsync or apt-get install rsync).

Common commands and typical use cases

scp examples

Copy a single file: scp /tmp/test.tar [email protected]:/tmp/ Copy an entire directory (recursive):

scp -r /local/dir [email protected]:/remote/dir

Download a remote file: scp [email protected]:/remote/file /local/path When to use scp :

Ad‑hoc transfer of small files where speed is not critical.

The target host does not have rsync and cannot install it.

rsync examples

Typical option set -avz: -a – archive mode (preserves permissions, timestamps, symlinks, etc.). -v – verbose output. -z – compress data during transfer.

Copy a single file:

rsync -avz /local/file [email protected]:/remote/path

Synchronize a directory (note the trailing slash):

# Sync the directory itself (including its name)
rsync -avz /data/app [email protected]:/data/

# Sync only the contents of the directory
rsync -avz /data/app/ [email protected]:/data/app/

Exclude files or directories:

rsync -avz --exclude=logs --exclude=*.tmp /data/app [email protected]:/data/

Resume an interrupted large‑file transfer:

rsync -avz --partial /data/bigfile.tar [email protected]:/data/

Limit bandwidth (e.g., 1 MiB/s):

rsync -avz --bwlimit=1024 /data/file [email protected]:/data/

When to use rsync :

Scheduled backups or periodic synchronizations (often combined with crontab).

Large files that change incrementally.

Need to preserve file attributes or exclude specific patterns.

Pitfalls to avoid

Trailing‑slash behavior – Adding / to the source path copies only the directory’s contents; omitting it copies the directory itself. Forgetting the slash can create an extra nested directory on the destination.

Permission handling with scp – Files copied with scp inherit the ownership of the remote user. Use rsync -a or adjust permissions manually after the copy.

Bandwidth throttling – For transfers over public networks, rsync --bwlimit helps avoid saturating the link.

Quick decision guide

scp : simple, full‑copy, ideal for occasional small files or when rsync is unavailable.

rsync : incremental, feature‑rich, best for regular synchronizations, large or frequently updated data, and when attribute preservation or exclusion is required.

Final recommendation

Use scp for quick, one‑off transfers of modest size. Choose rsync for batch synchronization, incremental updates, or scheduled backups to achieve higher efficiency and reliability.