PGP Whole Disk Encryption for Mac, benchmarks
:: Wednesday, October 1st, 2008 @ 12:56:04 am
:: Tags: Computers
Currently, most of my sensitive data is encrypted via encrypted sparse images, but I find managing encrypted volumes to be a pain in the butt (mount, unmount, mount, unmount).
I recently purchased PGP Desktop for Mac, which (finally) features pre-boot whole disk encryption. I love the idea that my drive(s) would be encrypted, and that a password would be required even before boot, but I wasn’t sure what kind of performance hit I’d get if I used PGP on an entire volume. And so, I ran some benchmarks.
The results aren’t pretty. I ran the benchmarks on a solid state drive hooked up through an SATA->IDE bridge (I know — I’m sorry. It was the only option, given my setup). At small transfer sizes, the PGP encrypted drive actually performed admirably, especially when compared to an unencrypted spinning drive. But disk performance overall takes a HUGE hit — in many cases, the drive performed 80% slower when PGP encrypted with default settings.
Assuming the machine was loaded up with enough RAM, I’ll bet most people wouldn’t even be able to tell the difference when doing normal tasks like emailing and web browsing. Anyone who isn’t computer savvy probably wouldn’t care, anyway. I find that many casual computer users just accept that computers are slow instead of taking action to rectify the situation (e.g. buying more RAM, taking the CD or DVD out of a Windoze machine, closing some applications).
As a photographer and media-heavy user, I have abandoned all hope of using PGP Whole Disk Encryption on my active OS drives. But I may end up using it on backup drives; portable drives are easier to steal than are notebook computers, and I don’t want my backup drives to be an easy way to get data from me.
And now, for the benchmarks:
Journal RSS 
Wow… those are some depressing numbers there.
[...] PGP Whole Disk Encryption for Mac, benchmarks [...]
So, Here’s some depressing numbers. before and after from QuickBench don’t think I can use WDE and get benefit from my X-25M
Standard Test Results (Cycles: 1) < Insert Chart Here > Seq. Read Seq. Write Ran. Read Ran. Write 4 KB 31.585 20.208 9.353 27.323 8 KB 58.744 45.162 20.763 27.078 16 KB 92.027 60.741 36.307 54.145 32 KB 112.122 29.952 63.771 19.485 64 KB 136.029 61.689 101.707 51.064 128 KB 169.068 67.01 140.681 44.506 256 KB 208.844 26.506 186.3 33.99 512 KB 229.571 57.076 214.202 49.319 1024 KB 243.366 51.743 233.922 42.058 Standard Ave 142.373 46.676 111.89 38.774
Standard Test Results (Cycles: 1) < Insert Chart Here > Seq. Read Seq. Write Ran. Read Ran. Write 4 KB 18.242 11.874 11.602 11.584 8 KB 23.325 20.067 14.99 8.954 16 KB 25.948 24.82 21.583 19.924 32 KB 28.316 26.754 24.557 15.061 64 KB 28.626 25.994 26.767 22.005 128 KB 29.789 21.122 28.318 12.667 256 KB 28.556 23.153 28.637 18.969 512 KB 30.612 23.109 28.664 18.175 1024 KB 30.517 23.287 28.507 18.57 Standard Ave 27.103 22.242 23.736 16.212
I wonder what difference the encryption settings make. Would the numbers look much better with 128bit encryption instead of 256? Some tests on this would be nice :-)