bitlocker performance hit??

Greetings,

Anybody have a feel for the performance overhead w/ bitlocker working, vs not?

New T61 2.4GHz Vista Ultimate, 3GB RAM 1GB turbo mem, 160GB encrypted HDD. Windows score 4.0.

Thanks BS

 

This is an interesting benchmark:

(Please do a web search for "Benchmark and Comparison of TrueCrypt and Bitlocker" on the ComeFrom Blog, and then replacing the "https" with "http" -- the forum rules prevent me from posting a URL, since I have fewer than 15 posts, so I need to give these convoluted instructions instead. Convenient, I know. :| )

It shows a substantial CPU load during the actual reading and writing (about 60% on one core on a 2 GHz dual-core system, 90% on one core and lower throughput on that same system underclocked to 1 GHz).

If these numbers reflect general performance values, I'd presume Microsoft's claim of a single-digit performance loss is a reference to "ordinary use", not specifically disk-intensive use. So a 0% loss during the 90% of the time when there's no disk activity, and a much bigger loss (for one core, anyway) during disk activity, averaging out to a single-digit hit overall.

 

I use BitLocker and, in everyday scenarios, the “single digit” performance hit is hardly noticeable if at all. This system still screams and the peace of mind knowing a thief will get nothing from my system if stolen is worth the negligible performance hit.

 

BinkNR, I think that for most users you're right.

I know I'm trying to figure out just this issue myself, since I intend to use my soon-to-be-here (but not nearly soon enough) Dell M6400 for a lot of data analysis, and I'm trying to figure out if working with extremely large data sets will imply a lot of disk activity, which in turn would make the Bitlocker hit more real....

The disk streaming test I referred to above, though, seems conceptually to be a worst-case scenario for Bitlocker. In random read/write tests, I bet the impact is much, much less; and in ordinary use (as you illustrate), less still.

 

I'm running a BitLocker on a Vista 64 2.4GHz 160GB X61 here, and I can't tell the difference during normal use.

I note that the Windows performance score is limited by the graphics (3.5), not by the disk data transfer rate, which actually gets the highest rating of any system component, even with BitLocker on (5.2).

 

Here is a proper link to a comparison of TrueCrypt and Bitlocker.

I am running Truecrypt on my x200 Tablet and don't see any measurable performance hit. Read and writes are solid and I don't think it is slowing down tasks (even disk intensive ones) by more than a few percentage points.

Graphics is also what holds my system back on the Vista Experience index. I have a 4.9 processor (1.86GHz LV C2D), 5.7 RAM (2GB DDR3), 3.9 graphics and 3.7 gaming graphics (Intel X4500 MHD), and a 5.3 hard disc (5400 RPM 320GB Hitachi 5k320).

Even if there is a small performance hit, I still recommend using it. I, and I assume many of you, take my ThinkPad everywhere and store a lot of data on it. I feel much better knowing that if I lose it, I am only out the price of the laptop, and don't have to worry about any of my data being compromised (you should also keep a backup on an external drive at home so you can get up and running quickly if you lose your laptop).

 

I am running BitLocker at my workplace notebook with Vista 64. I did not do any measurements (other than observing usage), but I hardly notice any difference running BitLocker. Maybe when writing lots of things it is a little bit slower, but I also think that graphics is holding me back more.

But I don't think you should compare Windows Experience Index hard drive scores, because my score did not change after applying BitLocker and running the test again, it still shows 5.2. My guess is that the test measures raw reading and writing capability of the drive, which is not affected by encryption.

 

I did some more thorough testing on TrueCrypt performance.

There does NOT SEEM TO BE ANY PERFORMANCE HIT for disk writes when using Truecrypt.

My x200 Tablet (1.86GHz C2D, 320GB Hitachi 5k320) has several partitions, some are encrypted and others are not. I tested my partitions individually using ATTO disk benchmark (tests reads and writes of multiple sizes).

My service partition (front of drive) is not encrypted and posted average read/writes of approximately 55/60 MB/s.

My Vista partion (immediately after service) is encrypted with TrueCrypt (AES). Read/writes were also approximately 55-60 MB/s. This is the same as the unencrypted partition. The only downside is increased processor cycles for the on the fly calculation (may reduce battery life slightly if you are doing disk intensive activity). Processor speed was not a buffer on my system, but if you have a low speed system with a fast disk (e.g. x300 w/ 1.2GHz C2D and Samsung SLC SSD) the encryption may slow you down by running out of processor.

My XP partition at the end of the drive (last 15GB) was much slower even though it is not encrypted. It only averaged a read write of about 27/31 MB/s.

Overall conclusion is that TrueCrypt (and presumably also bitlocker) does not directly reduce disk performance. However, if you are running a task that is both process AND disk intensive (e.g. compressing extremely high bitrate video), you may have a performance hit because cycles are being stolen from the application to encrypt the disk activity.

I have included screen grabs of my test results so you can examine them in detail and compare them to your own.