ASM over NFS or not?

Hello,
simple question: ASM over NFS (in particular NFSv3 on NetApp FS250) is faster than normal datafile on NFS or not?
Which is your experience?
Thanks
Stefano

Hi Stefano,
I have not seen such a configuration in production till now, but NetApp has a document about the different configurations and their performance:
http://www.netapp.com/library/tr/3423.pdf
Hope this helps you.
Best regards,
Matthias

Similar Messages

  • DNFS with ASM over dNFS with file system - advantages and disadvantages.

    Hello Experts,
    We are creating a 2-node RAC. There will be 3-4 DBs whose instances will be across these nodes.
    For storage we have 2 options - dNFS with ASM and dNFS without ASM.
    The advantages of ASM are well known --
    1. Easier administration for DBA, as using this 'layer', we know the storage very well.
    2. automatic re-balancing and dynamic reconfiguration.
    3. Stripping and mirroring (though we are not using this option in our env, external redundancy is provided at storage level).
    4. Less (or no) dependency on storage admin for DB file related tasks.
    5. Oracle also recommends to use ASM rather than file system storage.
    Advantages of DNFS(Direct Network File System) ---
    1. Oracle bypasses the OS layer, directly connects to storage.
    2. Better performance as user's data need not to be loaded in OS's kernel.
    3. It load balances across multiple network interfaces in a similar fashion to how ASM operates in SAN environments.
    Now if we combine these 2 options , how will be that configuration in terms of administration/manageability/performance/downtime in future in case of migration.
    I have collected some points.
    In favor of 'NOT' HAVING ASM--
    1. ASM is an extra layer on top of storage so if using dNFS ,this layer should be removed as there are no performance benefits.
    2. store the data in file system rather than ASM.
    3. Stripping will be provided  at storage level (not very much sure about this).
    4. External redundancy is being used at storage level so its better to remove ASM.
    points for 'HAVING' ASM with dNFS --
    1. If we remove ASM then DBA has no or very less control over storage. He can't even see how much is the free space left as physical level.
    2. Stripping option is there to gain performance benefits
    3. Multiplexing has benefits over mirroring when it comes to recovery.
    (e.g, suppose a database is created with only 1 controlfile as external mirroring is in place at storage level , and another database is created with 2 copies (multiplexed within Oracle level), and an rm command was issued to remove that file then definitely there will be a time difference between restoring the file back.)
    4. Now familiar and comfortable with ASM.
    I have checked MOS also but could not come to any conclusion, Oracle says --
    "Please also note that ASM is not required for using Direct NFS and NAS. ASM can be used if customers feel that ASM functionality is a value-add in their environment. " ------How to configure ASM on top of dNFS disks in 11gR2 (Doc ID 1570073.1)
    Kindly advise which one I should go with. I would love to go with ASM but If this turned out to be a wrong design in future, I want to make sure it is corrected in the first place itself.
    Regards,
    Hemant

    I agree, having ASM on NFS is going to give little benefit whilst adding complexity.  NAS will carrying out mirroring and stripping through hardware where as ASM using software.
    I would recommend DNFS only if NFS performance isn't acceptable as DNFS introduce an additional layer with potential bugs!  When I first used DNFS in 11gR1, I came across lots of bugs and worked with Oracle Support to have them all resolved.  I recommend having read of this metalink note:
    Required Diagnostic for Direct NFS Issues and Recommended Patches for 11.1.0.7 Version (Doc ID 840059.1)
    Most of the fixes have been rolled into 11gR2 and I'm not sure what the state of play is on 12c.
    Hope this helps
    ZedDBA

  • Incorrect file sizes shown in Finder over NFS and permissions issues

    Hi there
    This is a problem that existed for me in Leopard and has not been resolved in Snow Leopard.
    I have an XSan with a Leopard server sharing over NFS and AFP. When I connect from a Leopard or Snow Leopard client over NFS the file sizes in Finder are incorrectly displayed. My Tiger clients work perfectly.
    Also, although it says I have read write access to the files over NFS I cannot save over an existing file when I make changes to it, I instead have to create a new version of it and remove the old one.
    Check the link to show a grab of one of the folders in question, the upper window is what the NFS shows me, the lower AFP. If you Get Info on the files over either connection the byte count is identical.
    http://www.the-9000.com/images/finder_anomaly.tiff
    Any info would be greatly appreciated.

    This is a problem that existed for me in Leopard and has not been resolved in Snow Leopard.
    Have you filed a bug report with Apple?
    http://developer.apple.com/bugreporter/
    If not, there's less of a chance they'll know about it and help fix it for you.
    Do things look OK from the command line in Terminal?
    It would probably be useful to use a tool like Wireshark to check out what each protocol is sending over the wire. That could at least narrow it down to being a client or server issue.
    Thanks
    --macko

  • Write permission over nfs for transmission files

    Hi all,
    I have a strange problem and I hope you could help me.
    Context :
    I have a server with :
    transmission-daemon to remotely download torrent
    pyload to remotely download ddl
    nfs server to provide write access over lan for linux clients
    samba server to provide write access over lan for windows clients
    And of course, a client wich connect to nfs share to manage files .
    Problem :
    nfs clients have write permission over files created by pyload and basic users of servers but not on transmission files.
    otherwise, samba or local basic user have write access on all files (even transmission)
    Configuration :
    On the server, I have a storage mounted in /mnt/data binded to /srv/nfs4/.
    Transmission save files with mask 022 in directory :
    /mnt/data/downloads/
    Pyload with mask 022 in :
    /mnt/data/downloads/ddl/
    Nfs give access to /srv/nfs4 with wirte permission :
    /srv/nfs4/ 192.168.100.0/24(rw,no_subtree_check)
    Infos :
    With I know and I tested, I can tell you :
    I have 770 (and not 700) permissions on /mnt/data for local users on servers
    Permission files should not be the cause. nfs and smbd are owned by root so logically, both of them can write.
    Smb clients can write with every permissions (777,770,700) on files
    Nfs clients seems have permission of "other". No write with 700 and 770. But write access with 777
    pyload files are owned by pyload:pyload
    transmission files are owned by transmission:transmission
    basic users files are owned by [the_user]:users
    So the problem exists only for transmission (not pyload) owned files from nfs client (not samba) and files are the same permissions .. I don't understand.
    I will try with more verbose nfs server but I take every idea  
    Thanks you.
    Last edited by xp-1000 (2015-05-01 12:46:00)

    The problem could be related to Mavericks permission problems. You can try the Bridge fix:
    Open Finder > (go) Computer > Macintosh HD (for me) > Users > Click once on the Home icon (main admin) > Get Info >
    click on lock and authorise bottom right > click on cog wheel drop down > click on 'Apply to enclosed items' > OK

  • Error message: network copying over 31 characters not allowed, no ownership

    I have a MacBook Pro and a Mac Mini on a wired home network. The mini has an outboard USB drive mounted.
    I backup the MacBook Pro to that drive. I did this by turning Appleshare on on the Mini, and turning on personal file sharing. I then connect to it from the MacBook Pro by going to the Network prefs pane.
    I use Silverkeeper to backup. Until yesterday all of this worked fine. Yesterday I removed the outboard drive from the Mini and added two other USB drives. Both were new, and I formatted both as Mac OS extended (Journaled).
    Now, when I try to run Silverkeeper to backup to either of those new drives, I receive a message telling me that copying to the drive ("a Server (AppleShare) volume") may lose some functionality:
    - files over 2GB are not supported
    - Ownerships & Permissions not supported
    - Names over 31 chars not supported
    Hunh?
    I don't understand why this just started happening, but it looks like an OS issue. Anyone?

    Hmmm, with those limits, it seems that it thinks it's an HFS volume instead of HFS+, or HFS+ formatted under Mac OS 9/Mac OS 8 that only supports file sizes of up to 2 GB in size.
    And Anthony has some good advice/comments on the connection method, seems HFS+ via NFS has some of those limits.
    I use FW drives on most of my Networks, but some USB, and NDAS drives, and use Tri-Backup, never see this probem...
    http://www.tri-edre.com/english/index.html
    30 day free trial. Can't say enough good about Tri-Backup.

  • Sharing DVD over NFS - performance question

    Hi,
    I'm trying to share the DVD over NFS.
    I have updated /etc/rmmount.conf with the following line:
    share cdrom* -o ro,anon=0Everything works well, except the reading speed, which is quite slow.
    Copying a ~200MB file from DVD over NFS takes ~12 minutes.
    Copying the same file from DVD to the local filesystem takes ~2 minutes,
    and copying this file from the hard drive over NFS takes ~20 seconds.
    This makes me believe that the issue is in the interworking of the NFS and DVD-ROM, for (hypothetical) example,
    the disk can not spin up to the max speed due to the random access.
    Has anyone tried sharing the DVDs with good performance? Should I read up and tweak some NFS parameters or abandon the idea of achieving good performance?
    I'm using Solaris 10 Update 7.
    Thanks in advance for any suggestions.

    Thanks for the suggestion.
    I'm not sure if it's applicable for my issue. From what I've read, it may help if the same data is read over and over,
    but not with the reading speed of just inserted DVD disks which are not yet cached (which is the usage pattern I'm trying to optimize for).

  • Clusterware with ASM on NFS / NAS

    Are there notes / experiences on using an NFS / NAS that provides the "LUNs" or "disks" for ASM in Grid Infrastructure install ?
    Oracle GI would be installed on local nodes but the Shared Storage that I am looking for is NFS / NAS (e.g. NetApp). Presumably ASMLib might not be required (even if the nodes are Oracle / RedHat Linux) ?
    Hemant K Chitale

    Hi Hemant,
    Hemant K Chitale wrote:
    Modifying ASM_DISKSTRING via an ALTER SYSTEM command isn't an option until and unless you already have ASM installed and running. 11gR2 Clusteware installation attempts to install and configure ASM (at least for OCR and Voting Disks).
    So it becomes a "chicken and egg situation" !You can configure ASM_DISKSTRING using OUI at step "Create ASM Disk Group" on button "Change Discovery Path".
    With devices created (by "dd") on an NFS mountpoint, oracleasm doesn't CREATEDISK unless I use losetup. Fine, I had the same issue testing 10gR2 ASM (although I have been told that 11gR2 does not support losetup).I belive, we will not need use losetup anymore.
    I installed Oracle Clusterware using 2 Hosts without Storage System using NFS. (it's for test only)
    Creating zero-padded files.
    dd if=/dev/zero of=/oracle/asm/devices/nfsdisk1 bs=8192 count=131072
    131072+0 records in
    131072+0 records out
    1073741824 bytes (1.1 GB) copied, 27.9333 seconds, 38.4 MB/s
    dd if=/dev/zero of=/oracle/asm/devices/nfsdisk2 bs=8192 count=131072
    131072+0 records in
    131072+0 records out
    1073741824 bytes (1.1 GB) copied, 19.9333 seconds, 50.4 MB/s
    dd if=/dev/zero of=/oracle/asm/devices/nfsdisk3 bs=8192 count=131072
    131072+0 records in
    131072+0 records out
    1073741824 bytes (1.1 GB) copied, 17.9333 seconds, 52.4 MB/s
    #  chown oracle.asmadmin /oracle/asm/devices/nfsdisk*
    #  chmod 0660 /oracle/asm/devices/nfsdisk*Configuring NFS
    ### NFS Server
    # cat /etc/exports
    /oracle/asm/devices     129.10.10.0/24(rw,sync,no_wdelay,insecure_locks,no_root_squash)
    ### Mounting NFS on both servers "/etc/fstab"
    server-nas:/oracle/asm/devices  /oracle/asm/disks  nfs  rw,bg,hard,nointr,tcp,vers=3,timeo=600,rsize=32768,wsize=32768,actimeo=0,acregmin=0,acregmax=0,acdirmin=0,acdirmax=0  0 0At step Create ASM Disk Group I used "Redundancy Normal" and changed "Change Discovery Path" to "/oracle/asm/disks/*"
    The pre-reqs of ASM Devices will fails, I checked "Ignore":
    Error Message:PRVF-7039 : File system exists on location "/oracle/asm/disks/nfsdisk1"
    Cause: Existing file system found on the specified location.
    Action: Ensure that the specified location does not have an existing file system.
    Error Message:PRVF-7039 : File system exists on location "/oracle/asm/disks/nfsdisk2"
    Cause: Existing file system found on the specified location.
    Action: Ensure that the specified location does not have an existing file system.
    Error Message:PRVF-7039 : File system exists on location "/oracle/asm/disks/nfsdisk3"
    Cause: Existing file system found on the specified location.
    Action: Ensure that the specified location does not have an existing file system.Output of root.sh.
    # NODE1
    ASM created and started successfully.
    Disk Group DATA created successfully.
    clscfg: -install mode specified
    Successfully accumulated necessary OCR keys.
    Creating OCR keys for user 'root', privgrp 'root'..
    Operation successful.
    CRS-4256: Updating the profile
    Successful addition of voting disk 948b607b2b554ffebf5d89ea1e68beff.
    Successful addition of voting disk 4ac78fefc7e94f78bf6749fa48edbe0e.
    Successful addition of voting disk 1a1931d2fcff4fa4bf9c8c4817c86b79.
    Successfully replaced voting disk group with +DATA.
    CRS-4256: Updating the profile
    CRS-4266: Voting file(s) successfully replaced
    ##  STATE    File Universal Id                File Name Disk group
    1. ONLINE   948b607b2b554ffebf5d89ea1e68beff (/oracle/asm/disks/nfsdisk1) [DATA]
    2. ONLINE   4ac78fefc7e94f78bf6749fa48edbe0e (/oracle/asm/disks/nfsdisk2) [DATA]
    3. ONLINE   1a1931d2fcff4fa4bf9c8c4817c86b79 (/oracle/asm/disks/nfsdisk3) [DATA]
    Located 3 voting disk(s).
    CRS-2672: Attempting to start 'ora.asm' on 'holanda'
    CRS-2676: Start of 'ora.asm' on 'holanda' succeeded
    CRS-2672: Attempting to start 'ora.DATA.dg' on 'holanda'
    CRS-2676: Start of 'ora.DATA.dg' on 'holanda' succeeded
    ACFS-9200: Supported
    ACFS-9200: Supported
    CRS-2672: Attempting to start 'ora.registry.acfs' on 'holanda'
    CRS-2676: Start of 'ora.registry.acfs' on 'holanda' succeeded
    Configure Oracle Grid Infrastructure for a Cluster ... succeededThe installation ended sucessfull and cluster is up.
    Regards,
    Levi Pereira

  • [resolved]mpd over nfs

    I just set up one of my computers as an NFS server, so that the other computer will have access to my music collection.  Once I got the /home/shared/music directory mounted on the client computer, on that computer, I used ncmpcpp to start a database update to add those remote files to the local mpd database.  It seems to be working- I can hear the drive reading on the server, and the database appears to be growing on the client.  My question is, this:  Is it actually sending every bit of my 32 gig musical collection over the net?  It didn't occur to me that this might happen until I noticed that it was taking an inordinate amount of time to update the database.
    My next question is Will this cause weird things to happen with my slocate and fam on the client computer?
    Last edited by Convergence (2009-02-25 03:21:40)

    Hm, your quesions are specific to your setup and difficult to answer (for me at least), but I'll try.. 
    First: you're mounting the share with a local address, right? (i.e. 192.168.*.*)  If that is the case then your music will only be sent over your LAN, not the internet. 
    Second: MPD should only be scanning the shared directory unless you've set some setting that I am unaware of to make a local copy of all your music.  It's just reading the tags of all your files, which over a LAN could dtake some time.
    Finally: slocate and fam should operate normally.  I used to do the same thing with my music collection and never ran into a problem.  If the share is mounted when an updatedb occures than it will include those files in the update for slocate.  Similarly FAM is designed to work fine with NFS shares, there shouldn't be any quirks.

  • Arch to Arch: NFS mounts not visible in file explorer

    Hi all,
    After a long use of Windows I'm coming back to Arch, but with some agony.
    I have a NAS which shares my home directory over both cifs and nfs. It runs ARCH.
    When I mount the share on my desktop (also Arch) I can browse to it in xterm. I can ls the directories and files without a problem. When I try to open them however, the terminal hangs. Copying, moving, opening, altering ... it all doesnt work. The Xterm just freezes.
    When i open the folder with pcmanfm, it doesn't even show the directories.
    The weird thing: yesterday it worked, today no go.
    Desktop fstab:
    192.168.1.2:/home/wouter /media/wouter nfs auto,user 0 0
    Server exports:
    /home/wouter 192.168.1.53(rw,async,nohide)
    - The NFS service are running fine because my ESXi server its datastore is stored over NFS on the NAS.
    - The NFS services on the desktop are running.
    - Both systems are running the latest package version.
    - File permissions are correctly set: 700 wouter:users

    Same problem here.

  • HT201317 When I connected my phone and my iPad not all the photos in my photo stream moved over. Why not and how can I fix it?

    When I connected my iPhone and iPad via iCloud not all photos in my photo stream copied over. Why not and how do I fix it?

    Welcome to the Apple community.
    Only the last 30 days of photos are kept in photo stream in the cloud for syncing. Devices that have been syncing with photo stream for some time may retain older photographs that newer devices do not receive from photo stream.

  • Over / Down Icons not in exported folder

    Over / Down Icons not in exported folder.
    I've created some action buttons, and when a user hovers over it turns blue, and when they click it turns red. The png files I used for this I put into a folder called "IMG" (configuratorname.assets/IMG/blue-50-26.png). This folder is located in the .assets directory that configurator created when I saved the .gpc file. When I export the panel, the custom png files (red & blue) aren't exported into the content folder. How do I get it to export the custom icon png files ? Also, do I have to use a custom path in the the Over Icon and Down Icon fields. Here is what is in there now "Locales/IMG/blue-50-26.png", I think somewhere I read that I need to use a ".." at the beginning so it's a relative reference, not absolute ? My goal is for this panel to work on various machines, and all the needed componets are locally resident and pointed to within the content folder.
    Thanks ahead of time for your help.
    One last question, should the png file that I'm using as a custom color for the buttons when clicked be the exact dimension of the button, or will it scale the png file ?

    In your case, the png files will not be export to content folder, these files will be exported to content/Configurator-Panel.assets/IMG/.
    You can not use ".." to set the path of icon pictures. The root directory for icon picture is configuratorname.assets/. The value for icon picture path are relative to this root directory. After exporting, the pictures will be exported to content/configuratorname.assets/.
    The png file be scale to fill the dimension of the button. We strongly suggest that you set the png file and the button in same size. If the size of png file is lager, it should be in same ratio with the size of the button.

  • 7410 NFS server not responding

    Greetings,
    Anyone seeing "NFS server ... not responding" from a client of a 7410?
    I have a 7410 with a single J4400 (22 1TB drives @ RAID1, 1 spare, 1 logzilla). It's running 2009.09.01.3.0,1-1.8, which I believe is the latest and greatest. There is one client, a T2000 running Solaris 10. We're using NFS v3 to mount four shares from the 7410. Mount options look like this:
    box-nge2:/export/oracle/data     -       /oradata/data   nfs     -   yes rw,bg,hard,nointr,rsize=32768,wsize=32768,proto=tcp,vers=3,noac,forcedirectioA fairly large (~1TB) Oracle database lives here (although there was very little activity in the database when the following occurred).
    Today I was cleaning up some old data files from a now unused Oracle instance. Pretty simple: rm -rf /oradata/data/DO-NOT-WANT/ . I was surprised to see the command appear to hang, and the "NFS server kwaltz-nge2 not responding still trying" message appear. Control-C eventually got me my prompt back.
    When the system became responsive again a few minutes later I tried deleting files one at a time. Deleting some "large" data files (i.e 200GB or more) was taking more than 2 minutes, and caused the NFS server not responding message. Smaller files would take a few seconds.
    Why is a simple "rm" command bringing the 7410 to its knees? Any thoughts? Thanks.
    Edited by: roymcmorran on Dec 23, 2009 3:44 PM

    After some update issues, we also lost NFS. I created a test share and exported it to one of our Solaris 10 hosts. I got an rpcbind failure. This failure wasn't corrected by restarting the NFS service, nor rebooting the 7410. Rather, I had to DISABLE the NFS service, then re-enable it. After that, all connectivity returned.
    Charles

  • Ierrs on 3C905B over NFS on solaris 8 x86

    I have 2 identical Dell PC's, one with Solaris 8 HW 07/01 and one with Solaris 2.6
    On the Solaris 8 box, when copying files to local disk over NFS, I get lots of ierrs
    on the elxl0 devices and the copy is very slow (NFS server is a netapp with gigabit). FTP
    shows no increase in ierrs. The 2.6 box is fine doing the same operation
    so I suspect Solaris 8. I've tried my install disks on a Dell pc with a 3C905C card
    and thats OK. Here is netstat -ain
    Name Mtu Net/Dest Address Ipkts Ierrs Opkts Oerrs Collis Queue
    lo0 8232 127.0.0.0 127.0.0.1 103 0 103 0 0 0
    lo0 8232 127.0.0.0 127.0.0.1 0 N/A 101 N/A N/A 0
    elxl0 1500 172.17.8.0 172.17.8.121 22768 4457 1995 0 0 0
    elxl0 1500 172.17.8.0 172.17.8.121 20542 N/A 1992 N/A N/A 0
    bash-2.03# netstat -k elxl0
    elxl0:
    opackets 2557 ipackets 30038 oerrors 0 ierrors 6016 collisions 0
    obytes 456750 rbytes 38685177 multixmt 0 multircv 79 brdcstxmt 10 brdcstrcv 3108
    unknowns 1908 blocked 0 ex_collisions 0 defer_xmts 77 align_errors 0
    fcs_errors 0 oflo 6012 uflo 0 runt_errors 0 missed 0 tx_late_collisions 0
    carrier_errors 1 noxmtbuf 0 norcvbuf 0 xmt_badinterp 0 rcv_badinterp 0
    intr 24005 xmtretry 0 opackets64 2557 ipackets64 30038 obytes64 456750
    rbytes64 38685177 ifspeed 100000000 duplex half media PHY/MII promisc off first_collisions 0
    multi_collisions 0 sqe_errors 0 macxmt_errors 0 toolong_errors 0
    Help!

    Hi Ithieme,
    Try with other media. This issue may be your media problem.
    Revert back.
    Thanks & Regards,
    Senthilkumar
    SUN - DTS

  • Is it possible to install a consumer app for mass usage that would alter the ios device proxy settings? Over the air not supervised. Can an app company use an MDM server to do this if the customer consents?

    Is it possible to install a consumer app for mass usage that would alter the ios device proxy settings? Over the air not supervised. Can an app company use an MDM server to do this if the customer consents?

    HI Chris,
    Thanks for taking the time to reply.
    Could you expand on your answer, which bit is very doubtful?
    Also in your knowledge would Apple work with a company and license it if the software they were developing was desired by consumer demand?
    Thanks
    Gary

  • Adobe Phoshop Elements 11 installed, restarted computer after install complete.  When launch, Elements Organizer licensing agreement pops up.  Accept and it continues to pop up..over and over.  Will not launch program.  Help?

    Adobe Phoshop Elements 11 installed, restarted computer after install complete.  When launch, Elements Organizer licensing agreement pops up.  Accept and it continues to pop up..over and over.  Will not launch program.  Help?

    Try launching the editor and entering your serial number there if asked.

Maybe you are looking for