Pilots and controllers RAID performance issues.

Hello world

I'll beging assuming that I am new to VMware.

I decided to build a server ESXi 4.1 autonomous based on map of Intel S5520HC.

http://www.Intel.com/products/server/motherboards/s5520hc/s5520hc-overview.htm

I used 5 Seagate 1 TB SATA HDD 7200 RPM 32 MB cache connected to the MB controller. Of course ESX couldn't see the RAID so I used this:

http://www.Intel.com/products/server/RAID-controllers/SROMBSASMR/SROMBSASMR-overview.htm

I connected the 4 hard drives in a RAID 10 conf and used this driver during installation so that this controller and R10 table to be found.

http://Downloadcenter.Intel.com/Detail_Desc.aspx?AGR=Y & DwnldID = 18453 & ProdId = 3148 & lang = eng & OSVersion = A %0 & DownloadType = drivers

I installed ESX on the 5th disc and also affected the ability of reserve as a "scratch" data store I configured ESX to use this data as a swap for virtual machines store. I also added a datastore iSCSI on a homebuild storage based on Openfiler using similar disks (RAID 10 Setup). All net connections are gigabit Intel on the 3Com gigabit switches. We have therefore 3 data warehouses. 1 is on one disc, the 2nd is on hardware RAID10, 3rd is on iSCSI gigabit. Reading and writing of the 2nd data store performance has been disastrous. It was much lower than these other two stores of data. (reading or writing would not exceed 20000 Kbps as seen on the performance of vSphere client tab.

I decided to swap with the controller on that is supported natively by ESX, so I used RS2BL080 Intel

http://www.Intel.com/products/server/RAID-controllers/RS2BL080/RS2BL080-overview.htm

so I used 4 same disks and RAID 10 configuration and performance was even worse.

I noticed that ESX used the same 1 controller driver since they are compatible. (megaraid sas)

I want to permanantly unload/remove this driver and let the ESX to use a native without reinstallation of course.

Any other suggestions to improve performance are welcome.

Thank you.

Welcome to the community,

It is not the driver, what questions it's the writing mode in which operates the RAID controller. RAID controllers usually work in write-through mode, unless they have a cache with battery attached backup module. With the cache with battery backup module, they are able to operate in mode of writeback who basically dope performance. ESX(I) entirely relies on the caching of the RAID controller and does not any software based in implementing caching for security reasons.

Generally, you will see rates of transfer between 5-20 MB/s in view write-through and > 80 MB / s in write-back mode.

André

Tags: VMware

Similar Questions

  • Fill and sign of performance issue

    I do the test for our business environment before deploy us our software users. Complete and sign is a feature that was for a long time and it's finally here. While I'm pleased with the performance of the new single CD drive and DC Acrobat opening and loading documents is much improved, but there is one area where I don't understand why there is a significant difference between DC Acrobat and Reader DC. I test using a Tablet Surface Pro with Windows 8.1 and it included the stylus. When you use the tool fill and sign, a drawing in Acrobat DC signature works beautiful. The signing of drawing out great and it's very natural to draw, sign and use handwriting. However, in CD player, using the same exact document, the signature of drawing has a terrible lag and missing stroke, so the signature has hardly any resemblance to what it is supposed to look like. It's that if you draw the signature really really slow give time to the software to catch curves that you draw with your tablet pen. It makes the fill feature and sign virtually unusable. I tried the same thing on my PC using the mouse of office and I noticed the same offset on CD, but on Acrobat DC there is no lag and works very well. I hope that the software engineering team is aware of this issue in the CD player and a solution is on the way, unless it is somehow a conspiracy to steer people away from the free reader software and force them to buy the license for Acrobat DC. Ironically, the web plugin to Adobe Reader of cloud doesn't have this problem of mismatch. I just thought it was strange. Anyone else noticing the same problem?

    Error of grammar in a sentence has been corrected.

    Well, the good news is that I've reproduced the problem on another machine, however you have noted that it is very strange where and under what conditions it can be reproduced, it can be a difficult challenge, but we're investigating it now.

    Thank you

    Josh

  • Pilots and controllers missing

    Hi all. My HP Pavilion g6 crashed a week ago and had to change the hard drive. Download Windows 7 Home Premium. Now my laptop is missing drivers & controllers. I suspect these d & c have been installed at the factory as I got no recovery disk when I bought the laptop. My laptop is running on a 64-bit system.

    The missing drivers & controllers...

    PCI device

    ID: PCI\VEN_10EC & DEV_5209 & SUBSYS_3567103C & REV_01

    Error code: 28

    SM Bus controller

    ID: PCI\VEN_1022 & DEV_780B & SUBSYS_3567103C & REV_13

    Error code: 28

    Standard-VGA-Grafikkarte

    ID: PCI\VEN_1002 & DEV_6760 & SUBSYS_3567103C & REV_00

    Error code: 10

    Please is - can someone HELP me with these questions or direct me to where I can get the missing drivers for the computer laptop brand and moddel.

    Thanks in advance!

    Hello:

    Install the AMD chipset driver and restart.  You want the 49.7 MB file.

    http://support.AMD.com/en-us/download/chipset?OS=Windows 7-64

    Graphics card:

    http://h10025.www1.HP.com/ewfrf/wc/softwareDownloadIndex?softwareitem=ob-103101-1&cc=us&DLC=en&LC=en&OS=4063&product=5198288&sw_lang=

    Card reader:

    http://h10025.www1.HP.com/ewfrf/wc/softwareDownloadIndex?softwareitem=ob-103100-1&cc=us&DLC=en&LC=en&OS=4063&product=5198288&sw_lang=

  • The value customer NLS_LANG AL32UTF8 and abnormal select performance issue

    Hello
    I use a simple statement via sqlplus select and noticed by passing the WE8ISO8859P15 NLS_LANG to AL32UTF8 execution of the query is 10 x more.


    DB: Oracle Database 10g Enterprise Edition Release 10.2.0.4.0 - Production 64-bit
    NLS_CHARACTERSET: AL32UTF8


    The table has 500,000 lines and some varchar2 (32), the number columns (10) and the date.

    The select statement has none where conditions and looks like:

    Select a, b, c, a_date, b_date, x c_date

    The NLS_LANG to the customer of the OIC is the only change before I log in to the instance through sqlplus to run the query.

    The runtime is abnormal and I have no idea since the client character set and server are all two AL32UTF8 and normally no conversion should occur.

    Thank you very much in advance for a clue.

    Best regards
    Andreas

    You might be right, I found this on metalink:

    Bug 6350579 - winding SQLPLUS with trimspool and linesize is slow
    DOC - ID: 6350579.8

    Werner

  • Performance issues at the opening of business rule

    Hello

    We work in collaboration with Hyperion version 9.2.1 and we have performance issues when opening the business rules. I have analyzed the issue and discovered that it has something to do with the assignment of access privileges to the rule.

    The authorisation scheme looks like as follows:

    User A is assigned to the group G1
    User B is assigned to the Group G2
    Group G1 is assigned to the XYZ group
    Group is assigned to the XYZ group G2
    Group XYZ is the owner of the 'basic user' delivery to the planning application.

    Without attributing any unrestricted access the rule company opens immediately.
    By assigning the privilege to access the group G1/G2 (post or get) the business rule opens immediately.
    By assigning access privilege to the XYZ group the business rule opens after 2-5 minutes.

    Anyone has an idea why this happens and how to fix this?

    Kind regards
    ULI

    Edited by: user13110201 the 12.05.2010 04:31

    This was a problem with the business rules for quite some time. Oracle made front and rear markets in later versions than yours; and they have issued patches address, if it is not completely solved the problem. Finally things seem to be much better in 11.1.1.3, although YMMV.

  • HP Elitebook 8570p - Windows XP USB very slow and the transfer rate performance issues.

    Hello. I was wondering if anyone else knows some performance issues with the HP Elitebook 8570p and Windows XP Pro - SP3.

    I get a VERY slow start, general poor performance (waiting time when opening files and programs).

    I asked to burn in tests that show that everything is good.

    USB transfer rates are VERY slow (tried all ports of a Seagate FreeAgent USB drive, which works fine on other laptops etc.) for example a 30-minute transfer a 7gig single file of the USB HD on the desktop. Only 3 minutes to do the same on another laptop.

    I think there are problems with the drivers, like Windows 7 works very well and does boot from a bootable USB.

    USB transfer rate are also good when starting from a Windows boot environment.

    I installed HP SoftPaq Download Manager, and all the drivers are up to date.

    Hello:

    I think outside of the box here... Select the Intel (r) 7 Series Chipset Family SATA AHCI Controller instead.

    If it does not, I raise the white flag.

    Paul

  • Smartphone blackBerry Storm performance issues (and suggestions)

    I want to start and say that I don't like my phone.  When it works (which is most of the time), it works fine.

    But there are certainly problems of software.  As a software engineer, the OS has some memory serious management problems.  This seems to cause most of the problems I'm having with the phone.  I'm running the latest Verizon S/W (v4.7.0.148).

    Problem #1:

    I am a paranoid person.  There are a lot of personal information on my phone, I have currently international calls helped that I need this feature.  So to protect me my phone is locked when it is the case or after a period of inactivity time.  It is a good practice.  Part of the draw to the fate of the Blackberry is the security that you can configure on the device (I'm not paranoid enough to allow all the protections of encryption - yet).  Problem of the storm is that it can take 1 to 5 minutes to get the phone unlocked when unholstering.  It is not an exaggeration.  As the performance of the memory gets worse, time takes more time.  If I had to make an emergency call, it could be the difference between life and death.  5 minutes is a long time to wait to call 911.  This must be corrected!

    Problem #2:

    The phone must be "reset" once a day to remember "lost."  It is just poor programming.  I don't know if this problem of BB OS or Apps I'm running, but in both cases, this isn't fair.  After the reboot if I go to the display of memory under options I could have about 3-5 MB of free memory.  After a few hours of use, this will almost always '0 '.  I installed MemoryUp Pro, which seems to have extended the period of time between the battery grips, but I still don't think we should ask all owners of BB Storm to pull there battery once a day.  The other answer I get is "do not run so many Apps.  I get all my applications when finished.  Are the only ones that I continue to turn: Blackberry Messenger, Yahoo Messenger, Google Talk and Google Maps (Latitude).  Other applications running, you can't get out (phone, browser).  I tried doesn't work is not the 'optional' tricks and performance has not really much better.

    Problem #3:

    Missed calls.  This is probably related to problems of memory above, but if the storm can be used as a phone, then I should buy something else.  Although the occurrence of this problem has gone down because I installed MemoryUp Pro and perform the reset once a day, sometimes still about 1 to 5 times.  It takes so long for the phone to 'wake up' when a call comes in (especially if it is holstered), at the time where I hit "Answer", a call has been Messaging voice or even worse, the phone doesn't ring at all.  All I get is a notification of appeal lacks the 'unknown number '.

    Problem #4:

    Battery performance could be improved.  The battery that comes with the storm has about 24 hours on 24 with the Radio and Bluetooth.  Really, this isn't a problem if you know it.  I solved the problem by buying a 2700 mAh battery for extended life of BoxWave I use when traveling.  Makes the storm bit bulky, it will give me 2-3 days of life.  I don't know how this could be improved if not

    I hope that the developers of RIM are trying to solve some of these performance issues.  Certain tasks should be a higher priority as they seem to have (like phone & Unlock).

    Sean

    Here is the link...

    http://www.BlackBerryForums.com/General-9500-series-discussion-storm/196496-latest-9530-OS-4-7-0-208...

  • Management performance issues and huge catalog

    I shoot architecture and real estate. I turn every day hundreds of frames and my current catalog becomes huge. I'm also starting to see a huge performance hit recently. Last night I just changed until 03:30 because LR would take some time as 20 seconds to perform a single task (like changing a slide or photo of the flag). A lot of time spinning beach ball. I have an external hard drive which I support to locally.

    What I was wondering is if I have everything except the most recent project files copy the disk external then keep it unplugged when I WIP edition if it would help performance issues?

    MacBook Pro (10.9.5) and LR CC running. I optimized my catalog and I also have a Cocktail for Mac running on my system to clean the memory, etc.

    Other ideas of huge catalogs management would be useful. I prefer not to make a new catalog for each customer that I shoot several each day and passing backward between catalog every day would be a pain in the thigh.

    Hi KristianWalker,

    Have you tried to increase/purge your Cache? See How to improve and accelerate the performance of Photoshop Lightroom for more ways to optimize Lightroom.

    Also what size is your catalog at the moment?

    Kind regards

    Assani

  • DB performance issue

    Hi friends,

    We know the performance issue with our applications/database oracle.

    I run the OEM and I got the following charts in the report:
    http://farm3.static.Flickr.com/2447/3613769336_1b142c9dd.jpg?v=0
    http://farm4.static.Flickr.com/3411/3612950303_1f83a9f20.jpg?v=0
    http://farm4.static.Flickr.com/3411/3612950303_1f83a9f20.jpg?v=0

    Is there evidence that these cards can give re: performance problem?
    What other charts in OEM who can help solve or give assistance performance problem?

    Thank you very much in advance

    ytterp2009 wrote:
    Thank you Charles,.

    The server has 16 GB of Ram... way means that, due to limitations of 32-bit on windows only 4 GB of Ram was used and the 12 remaining GB is inactive?
    What is the proper way to tune?

    There are restrictions on how the memory beyond 4 GB can be used by Oracle. As far as I know, the memory is reserved for the buffer cache.

    Not all memory below the 4 GB limit is available for programs. Some of this memory is laid out in memory of the device, such as the video cards, RAID cache on board, etc., and some of the memory Windows is needed to track the objects window, icons, etc. Although it is possible to back up 2 GB per process limit to 3 GB, making Windows limit to use for its own purposes the memory that remains after the memory is drawn to embedded devices, and I think that the Oracle Metalink site does not recommend trying to go up the by the limit process completely to 3 GB. And this low by the process boundary should be reserved for the memory window. Your current value of pga_aggregate_target should be checked, because there is no way that the apparent present value could enter into the available memory.

    I agree with Martin - the best way to take full advantage of the memory is to pass on a 64-bit platform. Most, if not all the memory limit headaches disappear on a 64-bit platform. Of Windows 2003 Standard edition supports up to 32 GB of memory, this could be an option for you. It may or may not be a huge effort, but this means system loss of time (which can be considered a reason to call it a huge effort). You will probably find that the server, if less then three years, will support a 64 bit version of the operating system, but check the 64-bit versions of all the drivers are available and consider replacing hard disks containing the operating system if you need to return to the current operating system.

    Charles Hooper
    IT Manager/Oracle DBA
    K & M-making Machine, Inc.

  • How to monitor the TX and RX on PERFORMANCE for virtual network adapter BASP monitor?

    How to monitor the TX and RX on PERFORMANCE for virtual network adapter BASP monitor?

    I have a virtual network adapter that is created with Broadcom.

    This virtual interface named "BASP eCard" visible on the performance monitor in the category 'Rhythm Pipe' but not in the category "Network Interface".

    This is problematic because the Tx and Rx (bytes received/send in second) is available in "Network Interface" that show physical network cards only.

    Someone has an idea?

    Thank you

    This issue is beyond the scope of this site (for consumers) and to be sure, you get the best (and fastest) reply, we have to ask either on Technet (for IT Pro) or MSDN (for developers)
    *
  • Performance issues when coming out of sleep

    I recently built a gaming PC. On Christmas day, I finished putting the thing together. It has an Intel i7-4790 CPU running Intel HD Graphics, 8 GB of Corsair Vengeance RAM, an Asus Z97 - A Mobo and a 2 terabyte Seagate HDD. I installed Windows and put everything in place. Then a week later, my MSI GTX 970 arrived in the mail.

    So, I take the thing to sleep and decreases the performance of the whole thing. Videos YouTube stutter, simple as games fight of World of Tanks to run at its usual rate of 60 fps. To avoid this, I simply restart my comp and then it works fine. performance is back to normal.

    A few weeks, Windows gave me a message on one of the Nvidia drivers causing performance issues when you wake up from sleep. I don't think a lot of it and rejected it. I'll try to reinstall all my Nvidia drivers. I uninstall the Intel HD graphics driver a few days ago and has not changed anything. I would appreciate an answer to this problem because I would like to continue to use the "sleep mode" because it consumes less energy and prolongs the life of my case of LED. Thank you!

    In order to diagnose your problem, we need run Windows performance toolkit, the instructions that are in this wiki

    If you have any questions do not hesitate to ask

    Please run the trace when you encounter the problem
  • upgrade performance issues - see my profile

    Hello world

    I have suffered from some performance issues and would like to hear comments on my current set in place.

    USE

    I use InDesign, Photoshop and Illustrator also throughout the day.

    SETUP

    -Intel i7 - 3770K @ 3, 5 GHz

    -16 GB Crucial RAM

    -NVIDIA GeForce GTX 660

    -WDC WD1000DHTZ-04N21V0 1 to 10 k RPM 64 MB cache SATA drive at 6 GB/s (internal)

    I have not yet attributed this player for something specific

    -WDC WD5000HHTZ-04N21V1 500 GB 10 k RPM 64 MB Cache SATA drive with 6 GB/s (internal)

    It's my drive disk OS, applications, files of work and scratch

    -Software: Win7, WAA CC (not this latest version, there were issues, we had to go back to the previous version).

    WEAKNESSES

    1 RAM is small, I put at level 32 GB this week.

    2. video card is pretty low, but I rarely use first so maybe ok for now?

    3 discs: do you think that improving an SSD for OS and Adobe CC?

    I look forward to your opinions. Thanks in advance!

    Thanks for the information, Mohammad,

    What would you suggest for my existing 10 k TPM?

    I am considering a 256 GB SSD for my OS and applications. But think my scratch disk 500 GB and the 1 TB for my working files.

    Your thoughts?

  • Performance issues with large number of nodes

    I am creating an application to display graphics (large), for example:

    graph.png

    But I ran into some performance issues, even for a relatively small number of nodes in the scene graph (+-2000 in the picture above). The graph is built, step by step, adding circles and paths to a StackPane. Circles and paths can be semi-transparant. For a small number of nodes, I get a solid 60 FPS, but this decreases over time to about 5 frames per second. As soon as I stop adding new nodes, the framerate shoot again up to 60 images per second. The framerate drop even when all the nodes are outside the viewport.

    My questions are:

    * Is Platform.runLater () call to 2000 times a minute too?

    * This might just be a problem with my graphics card? (I have an Intel HD Graphics 3000)
    * JavaFX pulse recorder says such things, are there meaningful information in that I'm missing?

    PULSE: 1287 [163MS:321MS]

    T14 (0 + 0ms): col of CSS

    T14 (0 + 5ms): layout pass

    T14 (6 + 152ms): waiting for the minutes of the previous

    T14 (158 + 0ms): copy the State for graphic rendering

    T12 (159 + 0ms): dirty opts calculated

    T12: Path of the slow form for null

    T12 (159 + 160ms): painted

    T12 (319 + 2ms): Presentable.present

    T12 (321 + 0ms): completed the presentation of painter

    Counters:

    Background image of the region used cached: 14

    NGRegion renderBackgroundShape slow path: 1

    Nodes displayed: 1839

    Nodes visited during rendering: 1840

    Kind regards

    Yuri

    Basically, try some optimization of performance ranging from the simple to the complex.   Each of the changes below may provide you with an increase in performance.  Some will probably increase performance much more that others (depending on where is the real bottleneck).  I would probably start by replace the paths to the lines and reduce the number of Platform.runLater calls (as adding nodes that fall within the viewport can be difficult only).

    > The framerate drop even when all the nodes are outside the viewport.

    Place the nodes in the graph, which fall inside the viewport.

    > Is Platform.runLater () call to 2000 times a minute too?

    Yes, there is no reason to call it more than 60 times a minute when the framerate of JavaFX is capped at 60 frames per second by default.

    > This might just be a problem with my graphics card? (I have an Intel HD Graphics 3000)

    Yes, it's a graphics system relatively low-end.  But, look at the comment of developer below - your CPU and choice of graphic primitives can also affect rendering speed.

    > Adding circles and the paths to a StackPane

    You don't need a StackPane to this, a group is a container easier and probably better.

    > may be semi-transparant

    Remove transparency * may * cause acceleration.

    ----

    You run may be:

    https://JavaFX-JIRA.Kenai.com/browse/RT-20405 : improve the rendering of path performance

    Maybe if you use lines rather than the paths, performance might improve.

    A comment by a developer on this performance tweak application is:

    «It is quite normal for applications that use arbitrary paths (if the node path objects, SVGPath, polyline, or polygon) because these paths are rendered in software.» As card circle, Ellipse, line and Rectangle very primitive forms easily to the operations that can be performed entirely on the GPU, which makes them essentially cheap. There is no need to compare the rendering of a complicated shape for rendering of simple primitives for this reason. »

    ----

    Setting the cache indicators can help, but probably only if you animate nodes.

    ----

    Present level of detail of your graph functionality so a chart with zoom out is not make as many nodes as a part of zoomed in.

    ----

    You run any code important calculation on the JavaFX application thread that could stall it?

    Can make you available a ftom so that others can reproduce your problems?

  • Adobe Air performance issues?

    I have recently finished a huge project in Adobe Muse and I love the new sensitive design tools. However, as the project continued to develop, Muse has become increasingly slow to react and interact with the sensitive cursor was almost impossible. As someone who worked for Apple, I'd be the first person to point the finger my Mac and offer to buy a new computer, but it is a new, maxed-out MacBook Pro 15 inches. All other Adobe CC applications work without problem. When Muse was first published, as I remember, it was written on top of the Adobe AIR platform, instead of being natively written in C++. That this has something to do with performance issues that I see? If so, is there something I can do to improve performance, I see?

    Air is not basis for Muse more for a long time. Nativly revised muse was released in early 2014.

    In my view, the Muse team works actively on performance boosts, but I think that we should not expect any miracle: no other application I know not a calculate and display real-time changing dynamically and the same objects interactively from time on endless pages in response.

    In my opinion, the only way to significantly increase the speed would be gray on all goods during the move of the scrubber.

  • Is there a performance issue associated with SPARQL property path?

    Is there a performance issue associated with SPARQL property path?

    When we use paths property is our SPARQL queries, the queries never ends. I have a general problem, or do you have a problem with Jena and jena adapter version?

    We run Oracle 12 c. The Jena adapter has version 12.1.0.1.

    Thank you

    By Oyvind

    None of these requests ends. We have 300000 tuples in the model.

    SELECT subclass, superClass

    TABLE (SEM_MATCH)

    "SELECT? subclass? superclass WHERE {? rdfs:subClassOf + subclass? superclass} LIMIT 10'.

    SEM_Models ('isall1_x'),

    NULL, null, null, null,

    (= T, timeout = 100, qid = 123, STRICT_DEFAULT = F'));

    SELECT subclass, superClass

    TABLE (SEM_MATCH)

    "SELECT? subclass? superclass WHERE {graph? g {? subclass rdfs:subClassOf +? superclass}} LIMIT 10'.

    SEM_Models ('isall1_x'),

    NULL, null, null, null,

    (= T, timeout = 100, qid = 123, STRICT_DEFAULT = F'));

Maybe you are looking for