Full Disk Encryption Performance With Ubuntu 25.04 + Framework Laptop 13 Strix Point

Written by Michael Larabel in Storage on 2 May 2025 at 08:30 AM EDT. Page 2 of 5. 65 Comments.
SQLite benchmark with settings of Threads / Copies: 1. No Encryption was the fastest.
SQLite benchmark with settings of Threads / Copies: 2. No Encryption was the fastest.
SQLite benchmark with settings of Threads / Copies: 4. No Encryption was the fastest.

For the widely-used SQLite embedded database library by desktop apps and other software, within write-heavy workloads there can a minor impact to using full disk encryption... In most typical scenarios you won't notice any difference but if carrying out hundreds or thousands of write transactions at a time, it can add up, especially if dealing with multiple threads. But for the most part this SQLite benchmark can be treated as a synthetic test.

Flexible IO Tester benchmark with settings of Type: Random Read, Engine: IO_uring, Direct: No, Block Size: 4KB, Job Count: 1, Disk Target: Default Test Directory. No Encryption was the fastest.
Flexible IO Tester benchmark with settings of Type: Random Read, Engine: IO_uring, Direct: No, Block Size: 4KB, Job Count: 2, Disk Target: Default Test Directory. No Encryption was the fastest.
Flexible IO Tester benchmark with settings of Type: Random Read, Engine: Linux AIO, Direct: No, Block Size: 4KB, Job Count: 1, Disk Target: Default Test Directory. No Encryption was the fastest.
Flexible IO Tester benchmark with settings of Type: Random Read, Engine: Linux AIO, Direct: No, Block Size: 4KB, Job Count: 2, Disk Target: Default Test Directory. No Encryption was the fastest.

Similarly, if carrying out a lot of random reads and random writes with FIO, there is a significant difference to using full disk encryption but this is ultimately just a synthetic test. And even with these performance hits, full disk encryption is a worthwhile feature if storing any personal information on your laptop in the event it's lost or stolen.

Related Articles