5 Replies Latest reply: May 22, 2013 1:19 AM by Mihael RSS

    Configuring storage for ASM.

    just a DBA
      Please correct if this is not quite true.

      Many ASM benefits are active when storage is JBOD, without any hardware RAID implementations. Therefore we can configure each disk as one LUN and create diskgroup with many disks.
      However:

      1. Big part of storage price is the controller, therefore not using their features means to waste money.

      2. Hardware RAID is always faster then software RAID, therefore hardware mirroring will be faster than diskgroup with normal redundancy.

      3. Diskgroup with normal redundancy requires double space. Hardware RAID 5/6 is fast and doesn't affect write performance even with large amount of disks.

      4. ASM doesn't have "hot spare" feature, failed disks should be replaced manually. This can affect reliability of database.
        • 1. Re: Configuring storage for ASM.
          rarain
          Hi,

          Its not true ASM benefits are active when storage is JBOD...we can configure each disk as one LUN and create diskgroup with many disks whether it is JBOD Storage or RAID storage.

          1. Big part of storage price is the controller, therefore not using their features means to waste money.

          When you create Normal or High Redundancy diskgroup then better to attach each failure group with different controller so that even if controller is faulty it will impact only one failure group and diskgroup data will be still accessible.

          2.. Hardware RAID is always faster then software RAID, therefore hardware mirroring will be faster than diskgroup with normal redundancy.

          Totally agree...You don't need to mirror the data at diskgroup level if Hardware mirroring is already configured.

          3.Diskgroup with normal redundancy requires double space. Hardware RAID 5/6 is fast and doesn't affect write performance even with large amount of disks.
          True

          4. ASM doesn't have "hot spare" feature, failed disks should be replaced manually. This can affect reliability of database.
          True..

          Thanks
          • 2. Re: Configuring storage for ASM.
            Catch_22
            ASM is not a software RAID. It's it not RAID at all.

            ASM is a software based implementation to provide data redundancy, but it is not RAID and works on the basis of file extents (AU units) distributed on available free space of disk failure groups.

            ASM, unlike RAID solutions provides more configuration and space management flexibility. For instance you can add disks without having to rebuild your RAID configuration. From a performance, redundancy and space utilization aspect, multiple ASM failure groups (single or multiple disks) can provide a better value for money than any RAID solution, in particular for Oracle Database products.

            ASM is a complementary product and may use devices that are a hardware RAID for additional performance and disk level redundancy.

            To comment some of your statements:

            The performance of RAID 5/6 depends on your storage controller since it requires additional I/O. Write performance is usually bad, typically equal to a single drive (parity) and it provides very slow recovery in case of a disk failure.

            RAID 0+1 will give you the best performance, albeit being the most expensive.

            http://en.wikipedia.org/wiki/Standard_RAID_levels

            There is no need for a spare disk in ASM since ASM is not RAID as previously mentioned. Please check the ASM concepts guide.
            • 3. Re: Configuring storage for ASM.
              just a DBA
              Thanks for answers.

              I will clarify the question. We already have IBM DS3500 storage with 24x300 Gb disks and SAS host interface. It has Dual Controller, therefore there is no need to configure failure groups depending on controller - in case of one controller failure, the other automatically takes ownership of its disks. How to configure the storage for ASM in order to achive best performance, availibility and space utilisation for Oracle database ?
              • 4. Re: Configuring storage for ASM.
                Billy~Verreynne
                just a DBA wrote:

                1. Big part of storage price is the controller, therefore not using their features means to waste money.
                That depends on the type of storage. We have 2 storage arrays, each with 24 SAS drive bays. With 2TB SAS drives, a single storage array will be 48TB in size... and the most expensive component of the storage array would be the disks. Not the storage array chassis. Not the high-speed PCI cards and cables. Not the storage array switch.
                2. Hardware RAID is always faster then software RAID, therefore hardware mirroring will be faster than diskgroup with normal redundancy.
                Not true anymore. RAID can be implemented in a variety of ways - and these days the performance of RAID depends on the implementation.
                3. Diskgroup with normal redundancy requires double space. Hardware RAID 5/6 is fast and doesn't affect write performance even with large amount of disks.
                Perhaps. Perhaps not. RAID 5/6 has overheads. And these could be mitigated by the implementation. Or not. Have a look at http://oaktable.net/content/baarf-battle-against-any-raid-5 and http://www.miracleas.com/BAARF/BAARF2.html

                 
                The problem with these statements you are making, is that these are made as absolute statements. Always the case. Always true (or always false). But in there are actually very few absolutes in information technology today.
                • 5. Re: Configuring storage for ASM.
                  Mihael
                  On hardware level you can configure pairs of disks as mirrors and add two hot spare disks. Then, on ASM level you can create diskgroup with external redundancy. In this case you will have ASM data balancing and hardware data protection.
                  If you need more space, then you can configure some disks as RAID5 and place there data that does not require fast write performance (arhivelogs, read-only data etc.)