Ubuntu Disk Encryption Benchmarks

[phoronix]

Phoronix: Ubuntu Disk Encryption Benchmarks

Introduced in Ubuntu 7.10 was install-time encryption support where using the alternate installer one can fully encrypt their disk in an LVM using dm-crypt. Unfortunately, the Ubiquity installer in Ubuntu 8.04 continues to lack LVM and encryption support, but using Ubuntu 8.04 Alpha 6 we have looked at the performance cost of this encrypted configuration on Ubuntu Linux. Rather than looking directly at the disk read/write overhead caused by the encryption process, we have provided some benchmarks to see how the real-world performance is impacted in both gaming and other desktop tasks.

http://www.phoronix.com/vr.php?view=12036

[Kano]

How about adding a simple comparison like:

hdparm -tT /dev/sda

for the real device and

hdparm -tT /dev/mapper/whatever

for the encypted one. That shows lowlevel read performance (if the enc. partition is not the first one use the correct partition for the first test).

[deanjo]

hdparm is useless as a benchmark. It's results fluctuate wildly. For example here is a few tests using hdparm, running it again right after a test completes.

/dev/md0:
Timing cached reads: 1270 MB in 2.00 seconds = 634.85 MB/sec
Timing buffered disk reads: 160 MB in 3.01 seconds = 53.10 MB/sec
/dev/md0:
Timing cached reads: 1652 MB in 2.00 seconds = 826.65 MB/sec
Timing buffered disk reads: 206 MB in 3.00 seconds = 68.60 MB/sec
/dev/md0:
Timing cached reads: 1934 MB in 2.00 seconds = 967.87 MB/sec
Timing buffered disk reads: 254 MB in 3.02 seconds = 84.15 MB/sec
/dev/md0:
Timing cached reads: 1696 MB in 2.00 seconds = 848.26 MB/sec
Timing buffered disk reads: 190 MB in 3.01 seconds = 63.04 MB/sec
/dev/md0:
Timing cached reads: 1270 MB in 2.00 seconds = 635.47 MB/sec
Timing buffered disk reads: 254 MB in 3.01 seconds = 84.43 MB/sec
/dev/md0:
Timing cached reads: 2004 MB in 2.00 seconds = 1002.70 MB/sec
Timing buffered disk reads: 242 MB in 3.01 seconds = 80.29 MB/sec
/dev/md0:
Timing cached reads: 1710 MB in 2.00 seconds = 855.65 MB/sec
Timing buffered disk reads: 216 MB in 3.00 seconds = 71.99 MB/sec


It can't be used at all to indicate real performance.

[Kano]

As soon as you have got no background tasks it is better. Use single or S bootoption.

[deanjo]

As soon as you have got no background tasks it is better. Use single or S bootoption.

That's how the tests were done.

The only difference was me leaning over and letting out some natual gas during the tests. So does that mean my bodily functions have a direct effect on drive performance?

[Kano]

Only small changes here:

for x in $(seq 3); do hdparm -tT /dev/sda; done

/dev/sda:
Timing cached reads: 1664 MB in 2.00 seconds = 832.53 MB/sec
Timing buffered disk reads: 244 MB in 3.02 seconds = 80.88 MB/sec

/dev/sda:
Timing cached reads: 1718 MB in 2.00 seconds = 859.67 MB/sec
Timing buffered disk reads: 240 MB in 3.02 seconds = 79.47 MB/sec

/dev/sda:
Timing cached reads: 1706 MB in 2.00 seconds = 852.88 MB/sec
Timing buffered disk reads: 242 MB in 3.01 seconds = 80.38 MB/sec

[hoho]

Really strange results you get, Danjo.
Mine are also pretty stable with ~1% fluctuations like Kano's.

[Ole-Martin Broz]

Good good, the impact is less than i really thought.

I just got myself a new box, which gets linux smashed in soon, maybe ill test encryption and check how the system gets affected.

Well, i dont got anything to hide, so no point for me, however, as i said, the impact was low,

Can some1 try this on a single core cpu ? i just got a feelin that makes the impact even bigger, everything does on single cores.( in windows especially)

[gemidjy]

Why are you putting the Phoronix sign on the images you publish on your site? It is not something that noone else can do, right? Even they are taken with a digicam, which puts you in even more miserable situation..

thanks in advance

[Michael]

Why are you putting the Phoronix sign on the images you publish on your site? It is not something that noone else can do, right? Even they are taken with a digicam, which puts you in even more miserable situation..

thanks in advance

It's quite simple actually... It's because of the unfortunate situation of "blog spammers" or whatever you want to call them. It's a large problem with those copying our content and posting it on their own website/blog in full without our written permission or even any attribution.... Watermarks are also needed for those hotlinking off of our servers.