Platform calculation results for Server and Server with viewing client

The calculation results for the Server or Server with viewing client contain information about the recommended volume of the disk subsystem, total streams and server specifications. To view the results of the platform calculation, select the type of calculation—Server or Server with viewing client (see Selecting the type of calculation):

Each time the streams are added to the list (see Selecting cameras), or the parameters affecting the calculation are changed (bit rate, streams recording, etc.), the parameters are recalculated.

Archive and disk for metadata size

The Total video archive size section shows both the size required for the video archive and the HDD marked capacity—the recommended hard drive size. In this case, it is taken into account that hard drive manufacturers use multiples of 1000 when designating capacity, while the real hard drive size is calculated in the binary system as a multiple of 1024. For example, if the HDD capacity marked by the manufacturer is 1 TB, then in fact 0,91 TB will be available for the video archive.

The Dedicated SSD for metadata and system log section shows the volume of the SSD required to store the metadata and system log. This information is displayed only for the Axxon One platform if data storage was enabled in the settings (see Archive and metadata storage parameters).

Total streams

The Total streams field contains the values for the total stream from IP cameras, stream for recording and stream transmitted to clients, calculated for the specified cameras:

Recommendations

Recommended platforms

The Recommended platforms tab of the Result of calculation section shows the results of platform calculation for processors with the selected architecture (see Selecting the CPU architecture):

The specification parameters are displayed in the following columns:

1. Number of servers—the number of servers recommended for the specified number of cameras.
2. Server platform—platform name and, if detection tools using the GPU are specified, links to GPU selection (see Types of detection tools). When you click on the link to the GPU selection, the documentation pages with the performance data will open.
3. Server CPU load—CPU load as a percentage.

4. Server RAM, GB—the recommended size and number of RAM modules for dual-channel operation, taking into account the number of processors on the platform. For example, “4 x 8GB” RAM means two 8GB RAM modules per processor for a dual-processor platform.
   

   Note

   RAM should be used in dual-channel mode or greater mode based on the number of processors in the platform. For example, at least 2 RAM modules must be installed for single-processor platforms, and at least 4 RAM modules for dual-processor platforms.

   Using RAM at a higher frequency, or using RAM in dual-channel mode or greater mode results in CPU load decrease, and therefore in the performance increase.

5. Server storage configuration—required configuration of the disk subsystem, in particular:
   1. Actual archive size and required HDD marked capacity  for storing the archive if No RAID calculation was chosen, or actual archive size if RAID level was chosen.
   2. Recommended RAID configuration if RAID level was selected:
      1. total number of disks per server,
      2. disk size,
      3. number of RAID groups,

      4. number of hotspare-disks.
         

         Example

         28x 4 TB (2x RAID-6 + 1x HS) means 2 RAID-6 groups with a total of 27 4 TB disks and 1 hotspare-disk.

   3. The SSD volume required to store metadata and system log.
6. Required storage recording rate—total recording rate which must be supported by the disk subsystem of the server.
   
7. LAN (in/out)—total rate of the video streams from IP cameras, to be provided by the bandwidth of the network interface of the server.

You can add an alternative platform to the calculation—for details, see Adding a platform for calculation by analogy.

Recommended GPU models

If one or more GPU-based detectors were selected for video analytics (see Types of detection tools (Video Analytics)), then recommended GPU models will be listed below the table with recommended platforms:

The calculation is based on the results of testing the corresponding GPUs (see GPU performance for Axxon One detection tools).

Features of filters displayed above the table:

1. GPU filter—this is the basic filter for graphics cards. You can enter the model name partially or fully, for example:
    
2. Custom filter allows you to select specific GPU models for different detection tools. To do so, set the required checkboxes in each detector's tab and click Apply Filters:
   
   
   Specifics:
   1. If all checkboxes for any detector are clear, it will be equivalent to selecting all checkboxes.
   2. To discard current changes, click the Cancel button.
   3. The Reset button restores the filter to its default state: all GPUs are selected for all detection tools.
3. Different GPU Models per combination limits the maximum number of unique GPU models in the recommendations. For example, if a 1 value is set, then no different GPU models are combined in each recommendation; while a 2 value allows combinations of up to two different GPU models in each recommendation. The possible range is from 1 to 5, a default value is 1.
   Specifics:
   1. If there are insufficient benchmark results for GPUs to generate recommendations, an empty table with the message "Unable to calculate. Try to increase number of different GPU models" is displayed. In such cases, it is recommended to increase the combination value in the filter.
   2. If the number of combinations is large, calculation time increases, resulting in an empty table with the message "The calculation will take a long time. Reduce the number of selected GPU models or the number of different GPU models." In such cases, it is recommended to decrease the combination value in the filter.

Features of the recommended GPU models table:

• Recommendations are grouped by the number of GPUs and their resource load parameters. Cards with the same resource load parameters are displayed within one group.
• You can expand each recommendation by clicking on it—a table with details will open.
• In the expanded table, the selected detection tools and their quantity are listed, along with GPU core load in percentage, decoding chip load (Nvidia NVDEC) in percentage, and the required video memory in megabytes.
• The TOTAL row shows the combined resource load values of the graphics cards.

In the example below, four Nvidia Quadro P2200 (5GB) cards are recommended, and two of them are grouped together (2x Nvidia Quadro P2200 (5GB)):

GPU selection recommendations

GPU requirements for configuring/viewing the cameras

In case of the Server without viewing, the recording takes place on the server, and the cameras are viewed on the remote operator workstations (Clients), therefore, connecting the monitor to the server may be required only for the initial setup and configuration of the system. For this purpose, any graphics card (including integrated into the motherboard) is suitable. There is only one requirement for Axxon One: the graphics card should support OpenGL version 2.0 (or higher) with the ARB_vertex_program, GL_EXT_blend_func_separate, and GL_ARB_framebuffer_object extensions.

For a Server with viewing, it is sufficient to use an NVIDIA GeForce GT520 / Intel HD Graphics 530 graphics card (integrated into a 6th generation Intel Core processor) or a higher-performance GPU.

For hardware accelerated video decoding (Intel Quick Sync), use the graphics card built into some Intel processors (for example, Intel HD Graphics 530 built into the 6th generation Intel Core processor). This technology allows for faster and more energy efficient processing of video content. For more information on Intel Quick Sync Technology, see the product documentation.

GPU/neural accelerator requirements for detection tools

For a server with detection tools that use GPU resources, the suitable models and the number of GPUs/accelerators should be selected according to the information from the GPU performance for detection tools section.

Sorting and filtration

Platforms can be filtered in the calculation results. For that, enter the search string in the Platforms filter field. The filter is not case-sensitive:

The platform calculation results in the Recommended platforms table can be sorted by any column using the  button in the header of the corresponding column.

To sort the results by multiple columns at once, select them with the Ctrl key pressed. A number indicating the column's order in sorting is displayed in each selected column.

For example, this screenshot shows the calculation results which are first sorted by the CPU load decrease, then by the RAM size decrease, and then by the number of servers decrease.

Import and export

Calculation results can be exported to an Excel file—for details, see Export calculation results to xls. Also system configuration selected in the process of calculating the platform can be saved to a file—for details, see Configuration import and export.