Oracle ASM on top of ZFS volumes

What kind of overhead is involved if we configure Oracle ASM on top of ZFS volumes? Has anyone performed benchmarks using this scheme?

I've not done it personally with DNFS but I have created a multitude of clusters with NetApp on their NFS file system without ever having an issue other than saving a lot of money.

Similar Messages

  • Oracle ASM Disk Name Lenght Limit

    Hi,
    Does anyone know wheter or not oracleasm has a maximum disk name lenght limit, I have a situation where "#oracleasm listdisks" shows the same name for 5 of my disks and seems to have truncated the last character from the disk name.
    So, now my ASM Diskgroup is not starting up because it sees that there are "missing" disks, but they are not missing, they are just listed with the same Disk Name.
    Commands used to create the disks:
    oracleasm createdisk RAC_DATA146_500GB_DISK001 /dev/emcpowerek1
    oracleasm createdisk RAC_DATA146_500GB_DISK002 /dev/emcpowerem1
    oracleasm createdisk RAC_DATA146_500GB_DISK003 /dev/emcpowerej1
    oracleasm createdisk RAC_DATA146_500GB_DISK004 /dev/emcpowerel1
    oracleasm createdisk RAC_DATA146_500GB_DISK005 /dev/emcpowerei1
    Now, see the output when I do a querydisk:
    [root@birac01 ~]# oracleasm querydisk /dev/emcpowerek1
    Disk "/dev/emcpowerek1" is marked an ASM disk with the label "RAC_DATA146_500GB_DISK00"
    [root@birac01 ~]# oracleasm querydisk /dev/emcpowerem1
    Disk "/dev/emcpowerem1" is marked an ASM disk with the label "RAC_DATA146_500GB_DISK00"
    [root@birac01 ~]# oracleasm querydisk /dev/emcpowerej1
    Disk "/dev/emcpowerej1" is marked an ASM disk with the label "RAC_DATA146_500GB_DISK00"
    [root@birac01 ~]# oracleasm querydisk /dev/emcpowerel1
    Disk "/dev/emcpowerel1" is marked an ASM disk with the label "RAC_DATA146_500GB_DISK00"
    [root@birac01 ~]# oracleasm querydisk /dev/emcpowerei1
    Disk "/dev/emcpowerei1" is marked an ASM disk with the label "RAC_DATA146_500GB_DISK00"
    So, it looks like the version of oracleasm that I'm running has a name lenght limit? IS this possible?
    [root@birac01 ~]# rpm -qa | grep oracleasm
    oracleasm-support-2.0.3-1
    oracleasm-2.6.9-55.ELsmp-2.0.3-1
    oracleasmlib-2.0.2-1
    Can anyone please help?
    Regards
    Stephan
    Edited by: user1675141 on Mar 22, 2011 2:10 AM

    Oracleasmlib does not seem to advertise such information. As such it might be a poor design that it is not verifiying user input.
    According to http://download.oracle.com/docs/cd/E11882_01/install.112/e17212/storage.htm#CWLIN294
    +"The disk names that you specify can contain uppercase letters, numbers, and the underscore character. They must start with an uppercase letter."+
    The Oracle ASM assistant documentation shows:
    +"A volume name of Oracle ASM can be a maximum of 11 alphanumeric characters; dashes are not allowed. The first character must be alphabetic"+
    Why are you using such long volume and device names in first place? I would abbreviate the names and make them shorter. The url above will also show you information about configuring multipath and udev.
    You could try to query device names using:
    /etc/init.d/oracleasm querydisk -p VOL1
    and rename volumes using:
    /etc/init.d/oracleasm renamedisk /dev/sdb1 VOL1
    Edited by: Dude on Mar 22, 2011 5:19 AM

  • ORACLE ASM OR ZFS IN SOLARIS 11

    ORACLE DBA EXPERTS
    What kind of storage technology is recommended in order to implement an Oracle 11gr2 database in a Solaris 11 OS?
    -Oracle ASM or the Solaris 11 ZFS filesystem.
    Its any difference between the performance of both storage technologies?
    Thanks for answer my questions and i wish to yours a very "Happy Hollydays".
    Sincerely,
    JOHN JAIRO GOMEZ LAVERDE

    From what I understand Oracle recommends ASM. You cannot directly compare ASM with ZFS. ASM is a storage solution providing data management, availability and redundancy specific to Oracle. ZFS is an advanced file system. Whether you choose one over the other will depend on your storage requirements and knowledge.

  • Oracle ASM slower?

    I setup two databases on the same server one configured to use Oracle ASM and the other just sitting on top of an LVM2 volume using the same set of disks. They are both configured exactly the same, but on the ASM database it takes 7 minutes to run a simple count on a large table and 2 minutes on the other. I noticed that when running the query on the ASM db the CPU usage stays very low, but on the other it goes to 100%. Is there some sort of tuning I need to do on the ASM side?

    Am using ASM extensively.. and no, it is not slower. If it seems to be, you need to pop the hood, inspect the setup and configuration and implementation - as it is likely flawed.
    On Linux, if you want to use ASM properly, then assign block devices directly to ASM. Do not put LVMs and cooked file systems and other software layers in between.
    From a Linux perspective: the only software layer that, IMO, can be below ASM is multipath. And if you're not using HBAs and external storage, or srp devices and Infiniband storage, then multipath is not needed and ASM should deal directly with the disks as block devices.

  • Oracle ASM rebalancing does change rowid??

    Hi, all.
    The db is 11.2.0.3 on a linux machine.
    I use ASM for storage management.
    When oracle ASM rebalancing occurs, does the rowid of rows change?
    In my production environment, some application modules depend on rowid of rows.
    Have a wonderful day.
    Best Regards.

    Hi
    When oracle ASM rebalancing occurs, does the rowid of rows change?No, the load balancing is doie by moving ASM allocation units i.e datafiles extents (synonymous to OS blocks) to other physical volumns within same diskgroup to make data spread across all volumes in a balanced way. It has nothing to do with rowids and you should not worry about it.
    Salman

  • Oracle ASM disk to WWID

    Anyone know of a way to take an Oracle ASM disk (i.e. oracle ORCL:TEST_ASMDATA006) and finding the wwid of the underlying LUN? We are running 11.2.0.3 EE on redhat 5.8 with asmlib installed. Our current thought is we can take major minor and back track to multipath.conf file which we manually maintain when we add a lun. Would prefer a more direct way to do this if there is one.
    Thanks
    Steve

    You can query the physical device of a ASMLib volume using:
    /etc/init.d/oracleasm querydisk -p ASMVOL
    Then query the WWID:
    scsi_id -g -u -s /block/sda

  • Oracle ASM doesn't like my Xen xvd* devices...

    I have 3 VMs all running OEL, sitting on top of a physical host. On the physical host, I have configured Linux multipathing to approx. 20 SAN luns.
    I am presenting the multipath device files, the entities in /dev/mapper, up to each VM. However, when my colleague, the Oracle consultant, goes to make use of the /dev/xvd* files I present, Oracle ASM errors out and tells him, I believe, that the device is not a valid partition, or does not have partitions. Something along those lines.
    (Note: normally Xen prevents you from presenting a disk device read-write to multiple VMs, but I override this functionality with a '!' flag in the VM's config file).
    We've never had this problem when doing multipathing on physical servers. We just tell Oracle ASM about the files in /dev/mapper/, and it goes on its merry way.
    What, if anything, can I do to get Oracle ASM happy with the /dev/xvd* files in the guest OS? (The files that correspond to the files in /dev/mapper/* on the physical OS.)

    user9095995 wrote:
    I am presenting the multipath device files, the entities in /dev/mapper, up to each VM. However, when my colleague, the Oracle consultant, goes to make use of the /dev/xvd* files I present, Oracle ASM errors out and tells him, I believe, that the device is not a valid partition, or does not have partitions. You need to fisk the /dev/xvdX devices first to give them a valid partition table. Once you've done that, you can then expose them to ASM as /dev/xvdX1 for example.

  • Oracle ASM

    Hi,
    I have a new server Sunfire X4100 4x146G SAS disks running Solaris 10 X86 U4.
    1 pair of disks is Hardware mirrored for the OS. I created Oracle home on this partition and installed oracle 10g.
    I have been trying to decide if using Oracle ASM for the othere two disks is a good idea or not. Should I us ASM to create a disk group with the two remaining disks and run my production database on it or should I stick with manual storage management and raid those disks in hardware or even ZFS... ahh the choices?
    The white papers talk about databases with loads of disks and controllers, I am not at that level of hardware and expenditure, YET!
    Do ASM and OMF remove the need for me to manage the datafiles for the tablespaces?
    Do they handle the layout and optimization of the tables within a tablespace or is that still down to the admin to define the size of the extents during expansion etc?
    Can I run a standby database using non ASM storage from a database which uses ASM storage? If this is possible how are the datafiles handled on the standby database?
    I guess I could keep reading but I hoped some benevolent sole might already know the answers and save me days of reading!
    Finally, How reliable is ASM anybody care to comment? I know I can recover data using my existing OS level database files and ASM makes me uneasy...

    You really want to invest in http://www.oracle.com/technology/bookstore/index.html
    Bottom line: with ASM, the database still runs against RAW disk (physical or LVM) for best possible performance. However, instead of grabbing an entire disk for each tablespace, ASM holds the metadata that tells the database instance where to read and write for specific tablespaces. This allows ASM to coordinate things like striping in a way that is really beneficial for the database.
    The white papers talk about databases with loads of
    disks and controllers, I am not at that level of
    hardware and expenditure, YET!ASM allows you to grow to that when needed.
    >
    Do ASM and OMF remove the need for me to manage the
    datafiles for the tablespaces?Effectively yes. ... remove? how about minimize ...
    Do they handle the layout and optimization of the
    tables within a tablespace or is that still down to
    the admin to define the size of the extents during
    expansion etc?ASM uses the principals of SAME (Stripe and Mirror Everything) to minimize admin.
    >
    Can I run a standby database using non ASM storage
    from a database which uses ASM storage? If this is
    possible how are the datafiles handled on the standby
    database?Logical or physical standby. (I only have experience in the physical standby matching the storage technique of the prime. But I don't see why it should impact the physical standby to mix.)
    >
    I guess I could keep reading but I hoped some
    benevolent sole might already know the answers and
    save me days of reading!I know what you mean. The book mentioned above does consolidate a lot of the reading - worth the price and effort.
    >
    Finally, How reliable is ASM anybody care to comment?Very - you are still using RAW disk and Oracle has been doing that for decades.
    I know I can recover data using my existing OS level
    database files and ASM makes me uneasy...Not unusual.

  • Using a ZFS volume with encryption on as the Virtual disk in a Sparc VM

    Hi guys,
    Working my way through understanding how virtualization works in VM for Sparc. One question I thought of deals with having encryption turned on in a ZFS volume. If I create a volume with encryption ON, and then attach that volume as the sole virtual disk in a guest domain, will the virtualization still work? Has anyone else dealt with a scenario like this? I'm about to try it now, I'll report back with my findings.

    From my understanding, if you create a ZFS disk with encryption ON, mount it on Solaris, and after that present that to the VM, you should not have problems. The encryption will work between the volume and Solaris. To the guest vm, it should be transparent.

  • Oracle ASM Configuration on Solaris Cluster - Oracle 11.2.0.3

    Hi,
    I want some clarifications!
    I need to set Active and Passive Cluster settup on Solaris 10 SPARC Operating System, the HA software is Solaris Cluster and Oracle 11.2.0.3.
    1) I understand "Single instance Oracle ASM is not supported with Oracle 11g release 2" so we need to go for Clustered ASM - is it required to use RAC framework in this case?
    2) When i use the RAC framework, do i need to have license for RAC?
    Am new to Oracle, any help is appreciated.
    Regards,
    Shashank

    Hi,
    I want some clarifications!
    I need to set Active and Passive Cluster settup on Solaris 10 SPARC Operating System, the HA software is Solaris Cluster and Oracle 11.2.0.3.
    1) I understand "Single instance Oracle ASM is not supported with Oracle 11g release 2" so we need to go for Clustered ASM - is it required to use RAC framework in this case?
    2) When i use the RAC framework, do i need to have license for RAC?
    Am new to Oracle, any help is appreciated.
    Regards,
    Shashank

  • Add zfs volume to Solaris 8 branded zone

    Hi,
    I need to add a zfs volume to a Solaris 8 branded zone.
    Basically ive created the zvol and added the following to the zone configuration.
    # zonecfg -z test
    zonecfg:test> add device
    zonecfg:test:device> set match=/dev/zvol/dsk/sol8/vol
    zonecfg:test:device> end
    When I boot the zone it comes up ok but I am unable to see the device, nothing in format, /dev/dsk etc etc
    Ive also tried to setmatch to the raw device as well to no avail.
    Basically I have numerous zvols to add and dont really want a load of mount points from the global zone then lofs back to the local zone
    Any ideas please??
    Thanks...

    Thanks but that's why I created zfs volumes and newfs'ed them to create UFS and presented those to the zone.
    In the end I just create a script in /etc/rc2.d and mounted the filesystems in there.

  • Oracle ASM  installation in Solaris Cluster

    hello Experts,
    Could someone please tell me how to install Oracle ASM in Solaris Cluster and how to integrate it into the cluster resources.
    Details,
    2 Nodes (Pri & Sec) solaris 10 SPARC 64 bit OS
    solaris cluster 3.3 u5/11
    Thanks & Regards

    hi,
    pls take a look at tihs doc
    http://docs.oracle.com/cd/E18728_01/html/821-2678/gjcwv.html
    regards,

  • How to find memory taken by Oracle processes using top command

    I wanted to know how to find the memory taken by Oracle processes using top command. The output of the top command is as follows as an example:
    Mem: 16436072k total, 16215708k used, 220364k free, 395128k buffers
    Swap: 25165816k total, 1168288k used, 23997528k free, 13366520k cached
    PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
    27281 oraprod 15 0 6705m 1.3g 1.3g S 0.0 8.6 61:44.71 oracle
    27383 oraprod 15 0 6702m 1.2g 1.2g S 0.0 7.7 2:22.75 oracle
    5199 oraprod 16 0 6745m 1.1g 1.0g S 0.0 6.8 2:51.23 oracle
    The different Oracle processes could be Oracle database, Oracle listener, Oracle enterprise manager etc.
    I hope, my question is clear as to how to find the memory taken by a process using top command.
    Please revert with the reply to my query.
    Regards

    a short and correct answer would be: you can't.
    As oracle uses a fair amount of shared memory, and that shared memory is attached to most of the oracle processes, the same memory appears a number of times.
    you should rephrase your question / what is it you want to achieve?
    you can ask oracle how much memory is assigned to it, v$sga, v$sgastat, v$process (for PGA). moreover, you as DBA are the one who configured that.
    you can look at major an minor faults per process to see what is causing the paging (if you have any).
    you can also look at the /proc/pid/statm pseudofilesystem for your oracle pids.

  • Strange issue in Oracle ASM on Two node RAC where in one ASM node shows all diskgroup while other node shows  missing node.

    We have Oracle datbase 11gR1 in RAC node with Oracle ASM.Recently our database server got crashed and we are trying to restore back services.
    Using Snapshot technologyBusiness copy we had synced all our disk on storage level. Post this when we are trying to start ASM instance on node 1 it is coming and showing all diskgroups but on other node it is throwing errot with missing e diskgroup.
    ORA-15032: not all alterations performed
    ORA-15040: diskgroup is incomplete
    ORA-15042: ASM disk "5" is missing
    Expert please share your views.
    Thanks,
    Tushar

    The I/O fabric layer on the other node failed to mount all storage LUNs - resulting in ASM being unable to mount a diskgroup as there are missing disks in that group.
    Rebooting is exactly what could be needed to reset the h/w and infrastructure used by that node, in order for it to see all the storage disks again. As node 1 sees all storage disks (and is working), the disk itself on the storage system is intact and usable.
    What is the o/s? What is the fabric layer? What is used on o/s for dealing with the I/O fabric layer?

  • Oracle ASM disk group

    Hi,
    Having a multiple Oracle ASM disk group is more beneficial for Oracle database? if no why? if yes why?
    Any metalink note if have.
    thanks

    Any metalink note do you refer?What is your database version?
    Please see these docs.
    11g ASM New Feature [ID 429098.1]
    ASM 11g New Features - How ASM Disk Resync Works. [ID 466326.1]
    Thanks,
    Hussein

Maybe you are looking for