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
Design Guide—Flexible Computer Vision Solutions with Dell APEX Infrastructure
Executive summary Architecture concepts and requirements Solution architecture Validation
Validation overview
Performance results and findings
Conclusion References

Performance results and findings

Home > Workload Solutions > Computer Vision > Guides > Design Guide—Flexible Computer Vision Solutions with Dell APEX Infrastructure > Performance results and findings

Performance results and findings

  • Different tests were performed to validate the performance of the components of the Dell APEX solution for both VMS and CV workloads. The main difference between the solution described in this guide and the previous study is the speed of the CPUs in the test (2.6 GHz versus 3.0 GHz), so we needed to ensure no adverse effects of this change.

    Storage tests

    In the Dell APEX configuration, the PowerScale version is A300, whereas we previously tested the A3000. Because they are almost identical, no issues were expected between the two.

    To confirm, we traced the traffic from the recorder to the A300 PowerScale archive and did not observe any issues. The amount of video packet loss between the systems was zero. We also observed that the amount of data written to the A300 was consistent with what was coming into the recorder from the cameras.

    VMS Tests

    This test looked at the Recorder performance because that is CPU sensitive when writing to the long-term store.

    We observed that the recorder VM CPU periodically spikes to 80% when writing video to the archive, but this was not a concern for the overall performance of the system.

    Computer Vision tests

    In this test, 70 cameras were simulated using Milestone and connected to the Ipsotek CV system.

    Max Camera Test

    This test enables cameras in Ipsotek until no more can be enabled and handled by the system. This test ran on a single processing node that has an A16 which appears to the ESXi operating system as 4 x A2 GPU cards. With the previous sports and entertainment test, 60 cameras were concurrently processed and supported with this configuration.

    APEX CPU and GPU usage

    The CPU and GPU metrics were gathered for the Dell APEX platform and look as follows:

    The metrics obtained indicate no issues because the CPU is <70% utilization and the GPU is consistent with previous validations.

    Dell APEX Private Cloud and VxRail CPU comparison

    This is the core test because the CPU on APEX is slightly less powerful (even though more cores are available).  It was found that with this particular configuration, the CPU runs with higher utilization on the APEX platform. Keep in mind that other configurations are available that have fewer cores and higher CPU clock frequency.

    As shown here, the Dell APEX Private Cloud node CPU is running ~5% higher utilization when compared to the 3.0 GHz ones used in past validations. This is not an issue because the max is still well within the performance capacity of the system with ~70% utilization.