«
»

HP’s E200 controller really sucks

7. November, 2009

A long time ago, i wrote a review of the HP ML110. In the comments, Paul indicated that the Performance of the E200 controllers was pretty bad, and i promised i would do benchmarks of that. Now we have a year later, and i indeed finally got the time and did those benchmarks.

For the benchmarks, i’ve used the free version of HDtune. I’ve benchmarked four systems, and five different disk configurations. Note that the free version only does benchmarks for disk reads, and it’s a not a very pervasive test. None of these benchmarks are scientific. They should serve as a general indicator of performance, not as a final world on this topic. I don’t have that much clue about benchmarking.

The first system is my computer at home: It has an i7-920 CPU at stock speed, with 3x2GB RAM at 1333 Mhz (which is a slight overclock, but within the spec of the memory i purchased). Attached to it’s ICH10R controller are an Intel X25-M G2 160GB (Firmware 02HA) and a WD1001FALS (1TB, 7×24), running Windows 7 x64.

The next system is my work laptop, which is a ThinkPad W500 with a 2.53 Ghz T9400 C2D CPU, with 4GB of RAM. Attached to it’s onboard controller is an OCZ Vertex 120GB (Firmware 1.40), running Windows 7 x64.

The third system is our Exchange Edge server, on which i dared to install a benchmark utility. It’s an IBM x3250 with two 70GB 15kRPM 2.5″ SAS drives installed, attached to an onboard LSI1064E SAS controller. The system has a Xeon 3040 2.4Ghz Dualcore CPU and 5 GB RAM. It is running Windows Server 2008 x64 SP2.

And the final system is a HP ML110 G5 with a 2.33 Ghz Xeon 3065 CPU, 8GB of RAM and a E200 with the latest firmware (1.78). Attached to that are 4 WD1001FALS drives in a RAID10 configuration. The E200 has a backup battery and 128MB of cache installed. The system is running Windows Server 2008 R2.

Please note that none of these benchmarks are scientific. They were done on real systems, with workload minimized as much as possible, but virus scanners and other mandatory background applications active. Both the laptop and the desktop have not been formatted since Windows 7 RC was installed (i migrated to Windows 7 RTM using Windows.old), but the ML110 was freshly setup and the only application that’s been installed so far is the HP ACU and Forefront Client Security. The Exchange Edge server has been in use since May 2008. As such, the ML110 is the “cleanest” machine out of these four.

Intel’s X25-M G2 160GB on an ICH10R (AHCI Mode)

Intel X25-M G2 on an ICH10R

This is how a graph should look. It’s nice, it’s clean, it’s fast. Intel’s X25-M G2 shows how a modern SSD and storage subsystem should behave. Clean, predictable performance.

OCZ’s Vertex 160GB on an ICH7 (AHCI Mode)

OCZ Vertex 120GB on an ICH7

Here’s the OCZ Vertex. It’s running on a machine that’s a lot slower than the one the X25-M is attached to, and it’s storage controller is also quite a bit older. It still shows remarkably good performance. It should also be considered that this Vertex is quite a bit older – it was bought in May 09. It’s still very fast and responsive and a good SSD.

2x IBM’s 73GB 15kRPM 2.5″ SAS Disks on an LSI Logic 1064E SAS Controller

LSI Logic 1064E SAS Controller with 2x IBM 73GB 15kRPM SAS Disks in RAID1

As you can see, this is the performance you get from the server hard disks on an entry-level controller in an entry-level system. It’s not astonishing, but the performance is very well acceptable.

Western Digital’s 1001FALS 1TB on an ICH10R (AHCI Mode)

WD 1001FALS on an ICH10R

Here’s how the Western Digital disk behaves on a proper controller. Please note that this is a single disk, not part of a RAID array. The performance is quite good.

4x WD’s 1001FALS 1TB on an HP E200 in RAID10

HP E200 Controller with 4 WD1001FALS in RAID10

And here’s how it shouldn’t look. Compare this to the stand-alone disks above, which exhibits better performance. HP fucked up bad on this one, and there’s no fix in sight. Stay away from the E200.

And as a final word: I really don’t have much of a clue about benchmarking. If you see an obvious error here, please state what you think. If possible, i will try to correct it.

Update: As requested in the comments, i upgraded the E200 to Firmware 1.84 and redid the benchmark. It looks roughly the same.

HP E200 with Firmware 1.84

Category: Hardware  |  Comment (RSS)  |  Trackback

29 Comments

  1. Paul:

    …finally…. a benchmark. Thank you for that!

    BTW I’ve got more reason not to use the E200i:
    - in RAID5 mode the performance is even worse
    - on Win2k3 SBS you’ve got BSODs when producing heavy load (e.g. backing up or applying a service pack) –> fixed with the latest driver (but that fix took 2 YEARS to be done!!!)
    - my Win2k3-32 /w latest firmware and drivers (3x750GB RAID5 /w 1 HotSpare (drive 4)) simply froze when one of the 3 disks failed –> now how would the OS know about a failing RAID member? A mature controller never let that information slip through (except the driver does syslog-entries/alerts/whatsoever). So when the OS froze whos fault is that?
    - when replacing the faulted drive (drive 2) the new drive wasn’t taken as new HotSpare but instead the whole servers performance went to almost zero whilst E200 rebuilt the original array on drive 1-3 and re-setup drive 4 as HotSpare. There might be reasons to do so (I am aware of none) but I think I should have been asked instead “Do you want to use (drive 2) as HotSpare or to rebuild the original array?”.

    Paul

    9. November, 2009, 18:06

  2. Bjoern:

    Hi Lukas,

    the Firmware-Version is pretty old, isn’t?
    Version 1.84 is the latest one, so could you re-test it and post the results?

    Thanks, Bjoern

    10. November, 2009, 11:09

  3. Lukas Beeler:

    Hi Bjoern,

    The firmware version i tested with is 1.78, not 1.72 as i wrote. Sorry. Fixed that.

    I will retest with 1.84 as soon as i’m able to. Right now, i need the machine for our Exchange 2010 migration.

    10. November, 2009, 11:39

  4. Lukas Beeler:

    Hi Bjoern,

    I’ve completed our Exchange 2010 migration, updated the firmware, and retested. I don’t think it made a difference.

    14. November, 2009, 08:29

  5. jason wong:

    Did you run the HP Array Configuration software? I have a e200 and I believe a lot of performance features relating to cache are turned off by default on this controller unless turned on through the ACU.

    2. December, 2009, 22:39

  6. Simon:

    Hi Lukas,

    Have you experienced any issues with running the e200 controller in your ML110 G5 with a firmware version on the controller greater than v1.72? I have seen it mentioned in a few places on the web that the HP ML110 G5 stops seeing the e200 controller if you update it’s firmware past v1.72 and HP only show the v1.72 version as a firmware download on the ML110 G5 support and driver download page. If I read your comment above correct you were running v1.78 on the controller during your testing so what I have read may not be true?

    Cheers,

    Simon

    20. December, 2009, 14:31

  7. Lukas Beeler:

    Simon,

    I’ve no idea what you read, but the controller works, albeit with low performance.

    Here’s a download link from HP for 1.84:
    http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareIndex.jsp?lang=en&cc=us&prodNameId=3182559&prodTypeId=329290&prodSeriesId=1157688&swLang=8&taskId=135&swEnvOID=181

    Regards,

    Lukas

    20. December, 2009, 13:38

  8. Simon:

    Hi Lukas,

    Thanks for the reply. I’m guessing you’re updating the e200′s firmware via the Windows based Online ROM Flash component since the Proliant 100 series of servers can’t use the HP Firmware Maintenance CD? Have you come across or seen anyway of extracting the firmware bin file from these latest firmware updates (eg: v1.84) so that they can be applied to the old bootable USM memory stick method of upgrading the firmware? The reason I ask is that my ML110′s don’t always have Windows or Linux installed on them to peform the Online ROM Flash method (ie: VMware ESX). I have tried though am not seeing a firmware bin file with these newer firmware updates once they have been extracted – I’m just wondering if I’ve perhaps missed a way of doing this.

    All the best,

    Simon

    20. December, 2009, 13:45

  9. Lukas Beeler:

    Hi Simon,

    Sorry, we’re a Microsoft shop. But the E200 is supported by a bare Windows 2008 R2 DVD, you could try flashing the server using the Online ROM flasher from Windows PE (after booting it from USB or DVD)?

    Regards,

    Lukas

    20. December, 2009, 13:54

  10. Simon:

    Hi Lukas,

    It is so annoying HP don’t allow the HP Firmware Maintenance CD to work with the HP Proliant 100 series of servers. I created a BartPE CD though when running the Online ROM flasher from this it said it couldn’t detect an e200 controller. Am putting on a temporary install of W2K8 onto the server to get around this limitation. Was going to get disk stats from it using iometer once built up as a local VMare vmfs store if interested?

    Thanks,

    Simon

    20. December, 2009, 13:59

  11. Lukas Beeler:

    Hi Simon,

    BartPE is based on Windows XP or Server 2003. Neither of them ship with an E200 driver in-box, so you’d have to add that manually during the build stage.

    Benchmarks would be interesting. Paul (Comment #1 on this post) has the same issue, and i believe he is using ESX/ESXi as well.

    Regards,

    Lukas

    20. December, 2009, 14:02

  12. Simon:

    Hi Lukas,

    Good point re: drivers in BartPE – I should have thought of that. :)

    I’ll leave a link to my findings once complete. Will be interesting to see how the e200 will stack up, performance wise, against the onboard SATA controller.

    Cheers,

    Simon

    20. December, 2009, 14:12

  13. Paul:

    …there’s a new driver for Win2k3 at http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareIndex.jsp?lang=en&cc=uk&prodNameId=3577715&prodTypeId=15351&prodSeriesId=3577708&swLang=8&taskId=135&swEnvOID=1005

    They claim these improvements:
    “Added support for handling logical drive failure more gracefully. This improves the system responsiveness after logical drive failure.
    Increased per I/O max transfer size from 512K to 1M.”
    And these fixes:
    “Improved I/O distribution in greater than 4 logical drive configurations.”

    Do you think you could give it a try, Lucas? I have no spare (non-productive) ML110G5.

    Thanks, Paul

    5. January, 2010, 10:59

  14. Lukas Beeler:

    Hi Paul,

    I currently use my ML110 to test out DPM 2010 with WS08R2. Luckily, there’s also a new driver for WS08R2:
    http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?lang=en&cc=uk&prodTypeId=15351&prodSeriesId=3577708&prodNameId=3577715&swEnvOID=4064&swLang=8&mode=2&taskId=135&swItem=MTX-08615b21190547299c5bc68449

    I’ve installed it. The machine still boots, and here’s the result:

    http://projectdream.org/~lb/new_driver.png

    5. January, 2010, 12:25

  15. Jason404:

    Hmmn, even after reading this I hope to get good enough RAID-5 performance with a E200 using ESXi. I do not know of any other option that I can afford.

    Does anybody know what the maximum number of drives the E200 can use for RAID-5? Will the performance with 3 SATA drives really be worse than a single drive?

    13. January, 2010, 16:40

  16. Paul:

    Lukas, would you mind to delete my last comment?
    Thanks, Paul

    13. January, 2010, 17:40

  17. Paul:

    Jason,
    it’s worse than a single NOTEBOOK drive. Have a look here:
    http://projectdream.org/wordpress/2008/10/14/hp-ml110-g5/ *

    You’ve got bad luck /w E200i.
    Have you tried the internal (onboard) SATA RAID? This gives no RAID5 but probably better performance.

    HTH, Paul

    *
    find my comment:
    Paul:

    Lukas,
    you don’t work for HP, do you?
    The performance of the e200i controller is on subground level. There is not much about the SATA drives attached to it. Each of them allone has a much better performance.

    My 4GB Win2k3 server (4×750GB RAID5 + Spare) has 4x “Samsung Spinpoint F1 , 3.5?, 750 GB 7200rpm , 8.9ms , 32MB cache”. To copy a 3GB file on the RAID 5 takes full 6 minutes!!!
    My laptop uses a ST9250421ASG (Seagate Momentus, 2.5?, 250 GB 7200rpm, 16MB cache) drive. Copying the same file locally takes 3:30 minutes. On a power optimized laptop drive!

    Now tell me again about the SATA drives attached to the e200i…

    to 850cc:
    unlike workstation esxi always reserves the full disk space you assign. That might be 100GB and so it takes ages to “prepare” the virtual disks. Only solution is to exchange the e200i to sth. better.

    13. January, 2010, 17:41

  18. Lukas Beeler:

    Paul, Sure, i deleted it.

    Jason,

    Look at the benchmarks above. I have 4 WD1001FALS in RAID10 (the RAID Level which should give you the most performance, aside from RAID0). And the result is SLOWER than a SINGLE WD1001FALS! This should be very telling on how bad this thing is.

    RAID5 performance will probably be a A LOT WORSE than RAID10. There are plenty of other RAID controllers for similar prices. 3ware for examples has good controllers that work with Whitebox ESXi machines. Check the Whitebox HCL.

    13. January, 2010, 17:44

  19. Jason404:

    Thanks a lot for the replies. I am definitely not going to use RAID-5 with an E200 now.

    @Paul
    I actually have a ML115 G5, which has RAID-0, 1, 5 and 10 on the onboard nVidia controller, but I cannot use it with ESXi. I am using Hyper-V 2.0 at the moment, until I get a suitable hardware RAID controller.

    @Lukas
    Do you know if the performance is bad with RAID-1? I am thinking of forgetting about RAID-5, and sticking with RAID-1. The 2-port 3ware 9650SE-2LP looks good, but it is still a lot more expensive than an E200 (but I can see why they are so cheap now).

    14. January, 2010, 10:40

  20. Lukas Beeler:

    Jason404, Actually, i have a ML110 that uses RAID1.

    It has two 160GB HP Disks (WD RE2, i believe), and two WD1001FALS.

    Here are the benchmarks:

    160GB Disks: http://projectdream.org/~lb/HPE200RAID1WD160GB.png
    WD1001FALS: http://projectdream.org/~lb/HPe200RAID1WD1001FALS.png

    14. January, 2010, 11:40

  21. Alan Wright:

    Hi,

    Is the cache enabled on the e200. I believe it is disbled by default. if you download the array configuration utility (ACU) fro the HP site you can enable it through the gui interface.

    Let me know if it improves. I have a vsphere setup with a 64mb cache. I swapped the disks out installed windows 2008 so i could enable the cache. I am getting about 4mb/s. I can’t figure out why its so slow. I have ordered a 128mb BBW Kit from canada so i can enable raid 5 byuy another couple of disks and see if that helps any.

    25. January, 2010, 23:09

  22. Paul:

    Sure, Alan,

    thanks for your hint, but surely it’s enabled (128 MB, 50% read/50%write).
    Besides caching improves reading and especially “writing” of small files but not 3GB-Files.

    Paul

    26. January, 2010, 07:12

  23. Graeme:

    Also, have you checked to make sure that you’ve got the latest SAS/SATA driver from HP.com as well as the latest Storport.sys update from Microsoft? All three updates (Firmware, HP SAS, and MS Storport) work very closely together to ensure performance and stability.

    8. March, 2010, 04:22

  24. Paul:

    there is a new driver at
    http://h20000.www2.hp.com/bizsupport/TechSupport/SoftwareDescription.jsp?lang=en&cc=uk&prodTypeId=15351&prodSeriesId=3577708&swItem=MTX-6c98a488b5e844ac90c5e80ffb&prodNameId=3577715&swEnvOID=1005&swLang=8&taskId=135&mode=4&idx=0

    In “Enhancements” thew write: “… Increased per I/O max transfer size from 512K to 1M.” which might have a positive impact on performance.
    Were you able and willing to give it a shot, Lukas?

    Thank you very much,
    Paul

    13. March, 2010, 08:17

  25. Paul:

    BTW they claimed the same (512K to 1 M) in January…
    What’s interesting: The January-driver does not even appear in Revision History. Actually that quality of documentation fits the one of the E200…
    But it might be a hint that the January driver was no improvement and they don’t like to document that.

    Paul

    13. March, 2010, 08:28

  26. VMware ESX(i) and Solid State Drives (SSD) – Intel X25-M vs Samsung MMCRE64G5MXP-0VB | TechHead.co.uk:

    [...] As mentioned earlier I found that there was a massive difference in both read and write results when performing these benchmarking tests with the Samsung SSD connected to the E200 (located in the PCIe x8 bay) Smart Array controller compared to when performing the same tests with it connected to the onboard ICH9 based SATA controller.  I will posting another article that covers this as it would appear that the E200 controller really offers sub-standard (to put it politely) performance results with SATA based storage which is consistent with the findings of Lukas Beeler in his post here. [...]

    23. March, 2010, 00:05

  27. jaferrando:

    We are having a very serious issue with BL460c servers running ESXi 3.5 U4. There’s a problem in the vmware cciss driver with the E200i controller, apparently related to cache issues, that periodically ends up hanging the system.

    8. April, 2010, 16:50

  28. Kunal Kumar:

    Can e200 controller support two raid 1+0 configs?

    28. July, 2010, 05:26

  29. Paul:

    Sure. Half as many 1+0s as you have drives connected.

    30. July, 2010, 06:17

Leave a comment