InfoScale™ 9.0 SmartIO for Solid-State Drives Solutions Guide - Solaris

Last Published:
Product(s): InfoScale & Storage Foundation (9.0)
Platform: Solaris
  1. Introducing SFHA Solutions SmartIO
    1.  
      About SmartIO for solid-state drives
    2.  
      About SmartIO in an SFHA environment
    3.  
      About SmartIO in an Active/Active cluster environment
    4. About SmartIO in the Solaris virtualized environments
      1.  
        Performing live migration between LDOMs in the SmartIO environment
    5.  
      About the SmartIO caching profiler tool
  2. Using the SmartIO feature: use cases
    1. About SmartIO read caching for applications running on VxVM volumes
      1.  
        Required configuration for SmartIO read caching for VxVM volumes
      2.  
        Automatic caching for VxVM volumes
      3.  
        Setting up SmartIO read caching for VxVM volumes
      4.  
        Verifying the VxVM cache area and monitoring the caching
    2. About SmartIO read caching for applications running on VxFS file systems
      1.  
        Required configuration for SmartIO read caching for a VxFS file system
      2.  
        Automatic caching for VxFS file systems
      3.  
        Setting up SmartIO read caching for VxFS file systems
      4.  
        Verifying the VxFS cache area and monitoring the caching
      5.  
        Customizing the caching behavior
    3. About SmartIO caching on SSD devices exported by FSS
      1.  
        Status of cache areas when nodes leave or join the cluster
      2.  
        Setting up cache areas using SSDs exported by FSS
    4. About SmartIO write-back caching for applications running on VxFS file systems
      1.  
        Required configuration for SmartIO write-back caching for a VxFS file system
      2.  
        Setting up SmartIO write-back caching for VxFS file systems
      3.  
        Verifying the VxFS cache area and monitoring the caching (writeback mode)
    5. About multiple SmartIO cache areas for read and write-back caching on VxFS file systems
      1.  
        About the smartiocache option
      2.  
        Converting VxFS cache areas from one type to another
      3.  
        Setting up multiple cache areas on a system
      4.  
        Verifying the VxFS cache areas
    6. About SmartIO caching for Oracle databases on VxFS file systems
      1.  
        Prerequisites and configuration for using the SmartIO plug-in for Oracle
      2.  
        Setting default SmartIO caching policies for a database running on a VxFS file system
      3.  
        Setting SmartIO caching policies for database objects
      4.  
        Pinning and unpinning database objects
      5.  
        Enabling and disabling caching for the database
      6.  
        Listing cache policy details for the database
      7.  
        Listing cache statistics for the database
    7. About SmartIO caching for databases on VxVM volumes
      1.  
        Applying a SmartIO database caching template for a VxVM volume
  3. Administering SmartIO
    1.  
      Creating a cache area
    2.  
      Displaying information about a cache area
    3. Enabling or disabling caching for a data object
      1.  
        Enabling or disabling caching for a file system
      2.  
        Enabling or disabling caching for a data volume
    4.  
      Adding a device to the cache area
    5.  
      Pausing caching from a volume to a cache area
    6.  
      Removing a device from the cache area
    7.  
      Destroying a cache area
    8.  
      Setting the attributes of the VxVM cache area
    9.  
      Setting or changing the caching mode for a VxFS cache area
    10.  
      Flushing dirty data from a writeback cache area
    11.  
      Tuning the writeback caching
    12. Viewing the SmartIO cache statistics
      1.  
        Viewing the detailed caching stats for a VxVM cache area
      2.  
        Viewing the detailed caching stats for a VxFS cache area
  4. Troubleshooting and error handling
    1. Support for a persistent or 'warm' VxVM cache
      1.  
        Primary volume failure with a stale cache could cause possible data corruption
      2.  
        Migrating a cache during HA failover is not supported
    2.  
      Cache area is lost after a disk failure (3158482)
    3.  
      Cache is not online after a reboot
    4.  
      Recovering the write-back cache after a node failure
  5. Appendix A. Command reference
    1.  
      SmartIO command reference

Support for a persistent or 'warm' VxVM cache

A warm cache means that the contents of the cache remain persistent across planned reboots. By default, SmartIO does not provide a warm cache capability for VxVM caches. The cache area metadata is not flushed during a system shut down. The cache is invalidated after the system reboot, and whenever the volumes need to be restarted. This behavior is known as a 'cold' cache.

Veritas does not recommend that you configure a warm cache, because it may lead to data inconsistency. However, in some circumstances, when proper care is taken, it can be beneficial to configure a warm cache. For example, in the case of a controlled, planned reboot you may want to explicitly enable a warm cache. This feature enables you to flush the metadata to the cache, thus creating a warm cache.

If you enable a persistent or warm cache, VxVM detects and invalidates the persistent cache if the data volume was updated while the cache was offline. In clustered environments, if the cache area on a node is persisted during a planned shutdown on that node, and the data volume is updated on another node, the warm cache contents are treated as stale and persisted cached data is evicted.

To enable a warm cache for a planned reboot

  1. Before the system reboot, shut down all the applications.
  2. Create the warm cache by running the following command:
    # sfcache offline --flushmeta cachearea_name

    If desired, you can add this command to your shutdown script.

    You must run the above command before the application restart or system reboot. If not, the data in the cache is purged and the cache is not populated after the reboot.