Your Browser is Out of Date

Nytro.ai uses technology that works best in other browsers.
For a full experience use one of the browsers below
Cloudera Data Platform (CDP) 7.1.6 with PowerFlex and PowerScale
Revision history Executive Summary Product overview Solution overview CDP deployment Testing and Validation Conclusion Appendix A Solution components
Introduction
Hardware and Software components
Best practices
Appendix B References

Best practices

Home > Storage > PowerFlex > White Papers > Cloudera Data Platform (CDP) 7.1.6 with PowerFlex and PowerScale > Best practices

Best practices

  • This section outlines the performance best practices that are followed in this solution.

    PowerFlex rack

    PowerFlex rack best practices include:

    • The minimum number of nodes for production workloads is eight.
    • Dell Technologies recommends homogenous node types for predictable performance.
    • The maximum number of devices in a storage pool is 300.
    • Change the passwords for all default accounts.
    • Use secure communication - HTTPS (TCP port 443) - to remotely access PowerFlex nodes.
    • Ensure that PowerFlex rack is compliant to an RCM.

    PowerFlex

    PowerFlex best practices include:

    • Configure the high-performance profile for MDM, SDS, and SDC.
    • Disable Read Flash cache and Read RAM cache for all flash clusters.
    • Sometimes, increasing device queue length value to 256 can improve the I/O concurrency.

    See PowerFlex Tuning Parameters for more details on PowerFlex performance tuning and best practices.

    PowerFlex rack network

    See PowerFlex Networking Best Practices for detailed networking best practices.

    PowerScale

    See the Dell EMC PowerScale Hadoop Data Storage best practice guide for PowerScale performance tuning and best practices.

    Cloudera Data Platform

    CDP best practices include:

    Minimize anonymous page faults

    Minimize anonymous page faults by setting vm.swappiness=1, freeing from page cache before “swapping” application pages. This setting reduces the OOM-killer invocation.

    1. Edit /etc/sysctl.conf in your editor of choice, and add following line:

    vm.swappiness=1

    1. Then, run the following commands:

    # sysctl –p

    # sysctl –a|grep "vm.swappiness"

    Disable transparent huge-page compaction

    Add the following command to /etc/rc.local to ensure that transparent huge page compaction remains disabled across reboots:

    echo "never" >/sys/kernel/mm/redhat_transparent_hugepage/enabled

    Disable transparent huge-page defragmentation

    Add the following command to /etc/rc.local to ensure that defragmentation remains disabled across reboots:

    echo "never" > /sys/kernel/mm/redhat_transparent_hugepage/defrag

    Network

    The backend network between PowerFlex nodes and PowerScale should be 40 Gbps, with Jumbo Frames set (MTU=9162) for the NICs and the switch ports.