Expansion overview

Deployment options: Netezza Performance Server for Cloud Pak for Data System

Expand Netezza Performance Server and redistribute data offline with Netezza Performance Server 11.2.0.0 through 11.2.1.10. As of Netezza Performance Server 11.2.1.11, this functionality is available with the addition of the extra argument --full to the nzredrexpand command. For other options available with Netezza Performance Server 11.2.1.11, see Expanding Netezza Performance Server for Cloud Pak for Data System.

The newly added hardware is suggested to be of similar configuration to the existing hardware. The nodes must have the following:
  • The same number and capacity disks.
  • The same number of CPU cores.
In some cases, the capacity might be slightly different. For example, you can have disks of both 3.84 TB and 4 TB size in an enclosure.

These requirements ensure consistent redundancy characteristics and predictable query performance.

The expansion process consists of two steps.
  1. Provisioning the Cloud Pak for Data System hardware nodes with an automation script.
    The automation script:
    • Sets up personality of the SPU nodes.
    • Sets up BMC parameters.
    • Sets up switch VLAN.

    The new hardware must undergo the IBM Cloud Pak for Data System provisioning process. It is expected that the expansion is done in a multiple of eight nodes (two enclosures) to maintain resiliency characteristics. After this step, the ap nodes command shows the necessary personality for the newly expanded nodes.

  2. Expanding a Netezza Performance Server instance and redistributing tables by running the nzredrexpand command.

    The expansion process is automated by the nzredrexpand command. Expansion might take several hours to complete, depending on the amount of data on the system that needs to be redistributed. nzredrexpand produces an initial estimate of the total time that is necessary, and reports progress frequently as the command runs.

    Netezza Performance Server configuration, topology expansion, and table redistribution are handled automatically by the nzredrexpand tool. When the Cloud Pak for Data System node configuration step is completed, all that needs to be done is to power on the new nodes and run nzredrexpand.

    Automation overview
    1. Database access is locked.
    2. Netezza Performance Server database is stopped.
    3. A host backup is taken and the current configuration is saved.
    4. nzredrexpand gets the node list from Platform Manager's rest interface; updated config files are created.
    5. The nzstart -expand command is started. nzredrexpand waits for the system to come Online.
    6. When the system is Online, another host backup is taken.
    7. The nzredr -redistribute command is started; the actual data distribution begins.
    8. Run nz_redistribute -status to monitor progress of re-distribution status.
    9. Database access is re-enabled and the system is Online and operational.
    Netezza Performance Server configuration
    Netezza Performance Server configuration is handled automatically by the nzredrexpand script. nzredrexpand updates the following configuration files:
    • initTopology.cfg

      The initTopology.cfg file is updated to include the location of the newly added data slices on the new disks.

    • spaandslot.cfg

      Contains spa (mirror domain) and slot assignments for the new nodes.

    • bmc_map

      The bmc_map file is update to include the BMC IP addresses and Cloud Pak for Data System node names of the new nodes.

    Data slice expansion - topology and system management

    The new nodes that are provisioned for expansion are added to the topology under one or more new mirror domains along with new data slices. By adding enclosures in groups of two (groups of eight nodes), you ensure that mirror domains contain the wanted number of eight nodes per domain for optimal resiliency characteristics against node and disk failure.