RAID Array / Disk Performance with Bitlocker - unexpected results

%3CLINGO-SUB%20id%3D%22lingo-sub-2575633%22%20slang%3D%22en-US%22%3ERAID%20Array%20%2F%20Disk%20Performance%20with%20Bitlocker%20-%20unexpected%20results%3C%2FLINGO-SUB%3E%3CLINGO-BODY%20id%3D%22lingo-body-2575633%22%20slang%3D%22en-US%22%3E%3CP%3EI've%20doing%20some%20testing%20on%20a%20system%20to%20identify%20the%20impact%20that%20Bitlocker%20Encryption%20has%20on%20the%20read%2Fwrite%20performance%20and%20the%20results%20have%20been%20interesting%20to%20say%20the%20least%3B%20I'm%20hoping%20someone%20can%20help%20to%20explain%20my%20findings.%3C%2FP%3E%3CP%3E%26nbsp%3B%3C%2FP%3E%3CP%3EI'm%20using%20an%20entry%20level%20server%20with%20quad%20core%20Xeon%20E-2224%20CPU%20and%2016GB%20RAM%20which%20has%20Windows%20Sever%202016%20Standard%20installed%20-%20and%20it's%20a%20totally%20fresh%20Windows%20install%20with%20all%20of%20the%20latest%26nbsp%3Brelevant%20drivers%20are%20installed.%3C%2FP%3E%3CP%3E%26nbsp%3B%3C%2FP%3E%3CP%3EMy%20storage%20array%20is%20configured%20through%20the%20integrated%20controller%20on%20the%20motherboard%20and%20is%20across%204%20x%2010TB%20WD%20Enterprise%20disks%20configured%20as%20follows%3A%3C%2FP%3E%3CP%3E%26nbsp%3B%3C%2FP%3E%3CTABLE%3E%3CTBODY%3E%3CTR%3E%3CTD%3E%3CP%3ERAID%20Level%3C%2FP%3E%3C%2FTD%3E%3CTD%3E%3CP%3ERAID5%3C%2FP%3E%3C%2FTD%3E%3C%2FTR%3E%3CTR%3E%3CTD%3E%3CP%3ELegacy%20Disk%20Geometry%20(C%2FH%2FS)%3C%2FP%3E%3C%2FTD%3E%3CTD%3E%3CP%3E65535%2F255%2F32%3C%2FP%3E%3C%2FTD%3E%3C%2FTR%3E%3CTR%3E%3CTD%3E%3CP%3EStrip%20Size%3C%2FP%3E%3C%2FTD%3E%3CTD%3E%3CP%3E64%20KiB%3C%2FP%3E%3C%2FTD%3E%3C%2FTR%3E%3CTR%3E%3CTD%3E%3CP%3EFull%20Stripe%20Size%3C%2FP%3E%3C%2FTD%3E%3CTD%3E%3CP%3E192%20KiB%3C%2FP%3E%3C%2FTD%3E%3C%2FTR%3E%3CTR%3E%3CTD%3E%3CP%3EDisk%200%3C%2FP%3E%3C%2FTD%3E%3CTD%3E%3CP%3EWDC%20WD102KRYZ-0%3C%2FP%3E%3C%2FTD%3E%3C%2FTR%3E%3CTR%3E%3CTD%3E%3CP%3EDisk%201%3C%2FP%3E%3C%2FTD%3E%3CTD%3E%3CP%3EWDC%20WD102KRYZ-0%3C%2FP%3E%3C%2FTD%3E%3C%2FTR%3E%3CTR%3E%3CTD%3E%3CP%3EDisk%202%3C%2FP%3E%3C%2FTD%3E%3CTD%3E%3CP%3EWDC%20WD102KRYZ-0%3C%2FP%3E%3C%2FTD%3E%3C%2FTR%3E%3CTR%3E%3CTD%3E%3CP%3EDisk%203%3C%2FP%3E%3C%2FTD%3E%3CTD%3E%3CP%3EWDC%20WD102KRYZ-0%3C%2FP%3E%3C%2FTD%3E%3C%2FTR%3E%3C%2FTBODY%3E%3C%2FTABLE%3E%3CP%3E%26nbsp%3B%3C%2FP%3E%3CP%3EI've%20provisioned%20a%20single%20large%20logical%20volume%20and%20I've%20created%20two%20partitions%20for%20my%20testing%20as%20follows%3A%3C%2FP%3E%3CP%3E%26nbsp%3B%3C%2FP%3E%3CDIV%3E%3CTABLE%20border%3D%221%22%20width%3D%22273px%22%20cellspacing%3D%220%22%20cellpadding%3D%220%22%3E%3CTBODY%3E%3CTR%3E%3CTD%20width%3D%2252px%22%3E%3CP%3E%3CSPAN%3EDrive%3C%2FSPAN%3E%3C%2FP%3E%3C%2FTD%3E%3CTD%20width%3D%2274px%22%3E%3CP%3E%3CSPAN%3ECapacity%3C%2FSPAN%3E%3C%2FP%3E%3C%2FTD%3E%3CTD%20width%3D%2251px%22%3E%3CP%3E%3CSPAN%3EFile%20System%3C%2FSPAN%3E%3C%2FP%3E%3C%2FTD%3E%3CTD%20width%3D%2295px%22%3E%3CP%3E%3CSPAN%3EAllocation%20Unit%20Size%3C%2FSPAN%3E%3C%2FP%3E%3C%2FTD%3E%3C%2FTR%3E%3CTR%3E%3CTD%20width%3D%2252px%22%3E%3CP%3EY%3A%3C%2FP%3E%3C%2FTD%3E%3CTD%20width%3D%2274px%22%3E%3CP%3E100%20GB%3C%2FP%3E%3C%2FTD%3E%3CTD%20width%3D%2251px%22%3E%3CP%3ENTFS%3C%2FP%3E%3C%2FTD%3E%3CTD%20width%3D%2295px%22%3E%3CP%3E4096%20bytes%3C%2FP%3E%3C%2FTD%3E%3C%2FTR%3E%3CTR%3E%3CTD%20width%3D%2252px%22%3E%3CP%3EZ%3A%3C%2FP%3E%3C%2FTD%3E%3CTD%20width%3D%2274px%22%3E%3CP%3E25%2C600%20GB%3C%2FP%3E%3C%2FTD%3E%3CTD%20width%3D%2251px%22%3E%3CP%3ENTFS%3C%2FP%3E%3C%2FTD%3E%3CTD%20width%3D%2295px%22%3E%3CP%3E8192%20bytes%3C%2FP%3E%3C%2FTD%3E%3C%2FTR%3E%3C%2FTBODY%3E%3C%2FTABLE%3E%3CP%3E%26nbsp%3B%3C%2FP%3E%3CP%3EI%20ran%20an%20array%20of%20tests%20(results%20attached)%20on%20the%20volumes%20before%20encryption%2C%20then%20reformatted%20and%20encrypted%20the%20volumes%20and%20same%20tests%20again.%3C%2FP%3E%3CP%3E%26nbsp%3B%3C%2FP%3E%3CP%3EThe%20results%20have%20been%20very%20interesting%2C%20I%20was%20expecting%20to%20see%20a%20small%20performance%20hit%20across%20all%20tests%20due%20to%20Bitlocker%2C%20but%20what%20I%20found%20was%20that%20some%20tests%20had%20a%20drastic%20performance%20decrease%2C%20in%20some%20cases%20up%20to%2050%25%2C%20and%20surprisingly%20some%20tests%20showed%20an%20increase%20in%20performance%2C%20in%20one%20case%20an%20increase%20of%2060%25.%3C%2FP%3E%3CP%3E%26nbsp%3B%3C%2FP%3E%3CP%3ECan%20anyone%20help%20to%20explain%20this%3F%26nbsp%3B%20I'm%20thinking%20of%20running%20the%20tests%20again%20with%20another%20product%20to%20see%20if%20I%20get%20similar%20results%2C%20but%20what%20I'm%20using%20is%20fairly%20robust%2Freliable%20from%20my%20experience.%3C%2FP%3E%3CP%3E%26nbsp%3B%3C%2FP%3E%3CP%3EFindings%20are%20in%20the%20spreadsheet%20attached.%3C%2FP%3E%3C%2FDIV%3E%3C%2FLINGO-BODY%3E%3CLINGO-LABS%20id%3D%22lingo-labs-2575633%22%20slang%3D%22en-US%22%3E%3CLINGO-LABEL%3EBitLocker%3C%2FLINGO-LABEL%3E%3CLINGO-LABEL%3EDisk%3C%2FLINGO-LABEL%3E%3CLINGO-LABEL%3EEncryption%3C%2FLINGO-LABEL%3E%3CLINGO-LABEL%3EPerformance%3C%2FLINGO-LABEL%3E%3CLINGO-LABEL%3Eraid%3C%2FLINGO-LABEL%3E%3C%2FLINGO-LABS%3E
Occasional Visitor

I've doing some testing on a system to identify the impact that Bitlocker Encryption has on the read/write performance and the results have been interesting to say the least; I'm hoping someone can help to explain my findings.

 

I'm using an entry level server with quad core Xeon E-2224 CPU and 16GB RAM which has Windows Sever 2016 Standard installed - and it's a totally fresh Windows install with all of the latest relevant drivers are installed.

 

My storage array is configured through the integrated controller on the motherboard and is across 4 x 10TB WD Enterprise disks configured as follows:

 

RAID Level

RAID5

Legacy Disk Geometry (C/H/S)

65535/255/32

Strip Size

64 KiB

Full Stripe Size

192 KiB

Disk 0

WDC WD102KRYZ-0

Disk 1

WDC WD102KRYZ-0

Disk 2

WDC WD102KRYZ-0

Disk 3

WDC WD102KRYZ-0

 

I've provisioned a single large logical volume and I've created two partitions for my testing as follows:

 

Drive

Capacity

File System

Allocation Unit Size

Y:

100 GB

NTFS

4096 bytes

Z:

25,600 GB

NTFS

8192 bytes

 

I ran an array of tests (results attached) on the volumes before encryption, then reformatted and encrypted the volumes and same tests again.

 

The results have been very interesting, I was expecting to see a small performance hit across all tests due to Bitlocker, but what I found was that some tests had a drastic performance decrease, in some cases up to 50%, and surprisingly some tests showed an increase in performance, in one case an increase of 60%.

 

Can anyone help to explain this?  I'm thinking of running the tests again with another product to see if I get similar results, but what I'm using is fairly robust/reliable from my experience.

 

Findings are in the spreadsheet attached.

0 Replies
www.000webhost.com