Performance guidelines
There are two components to consider when looking at FortiRecorder performance – the FortiRecorder and the Client computer with FortiRecorder Central or a browser. Overall FortiRecorder performance is a combination of the video input (video compression, image quality level, complexity of the scene, video resolution, frame rate per second, number of cameras) and the video output (to the clients for live views and playback). The performance bottleneck in a FortiCamera deployment will likely be the network bandwidth to and from FortiRecorder and the CPU performance of the computer running the FortiRecorder Central or browser client, which must decode and render the video streams from the FortiRecorder. Displaying multiple video streams on the client is very CPU intensive.
FortiRecorder performance
Number of supported cameras
The FortiRecorder-200D and FortiRecorder-400D can support up to 64 cameras depending on the configuration. The FortiRecorder-100D is suitable for 16 cameras. For FortiRecorder-VM, the number of supported cameras is dependent on the hardware configuration of the VMware server and the number of licensed cameras.
General performance factors
The following factors affect the input side of performance:
- Total number of video streams from the cameras (i.e. not just the number of cameras).
- The video recording types (motion only or continuous) per camera.
- The video stream parameters per camera – i.e. resolution, frame rate, bitrate mode (constant or variable) and the bitrate mode parameters (bitrate or image quality)
- The number of detection events being received and the number of associated snapshots and clips generated for display in the event monitor.
- Storage settings - moving recordings from local to remote storage, NAS model and type, network connectivity to NAS, recompression and deleting of continuous recordings when detection recorders have to be kept.
The following factors affect the output side of performance:
- Number of administrator/operator/viewer sessions
- Peak number of simultaneous administrator/operator/viewer live views
- The video stream parameters per camera live view – i.e. resolution, frame rate, bitrate mode (constant or variable) and the bitrate mode parameters (bitrate or image quality).
Variable versus constant bit rate
The variable bit rate mode means the bandwidth used by the camera will vary according to what the camera is seeing and the video profile settings. The video profile settings for the variable bit rate mode are resolution, frame rate and image quality. High resolution creates more data than medium or low resolution (see following sections for more detail). The degree of motion present in a video stream also affects the amount of data created.
The constant bit rate mode means the bandwidth used by the camera will stay relatively constant regardless of what the camera is seeing. The constant bit rate mode is therefore more predictable in deployments where bandwidth and/or storage capacities are important considerations. The video profile settings for the constant bit rate mode are resolution, frame rate and bit rate. The bandwidth used by the stream is dictated by the bit rate setting.
In general, using the variable bit rate mode results in relatively consistent video quality but fluctuating bandwidth and using the constant bit rate mode results in varying video quality but predictable bandwidth. Choosing a high bandwidth constant bit rate mode avoids the video quality drop e.g. during high motion, but may use some unnecessary bandwidth during times of no activity.
However, in most cases the difference in video quality between the variable and constant bit modes is negligible (assuming the same resolution and frame rates) and the constant bit rate mode produces more reliable output from the cameras.
Bandwidth per camera
Variable bit rate
Depending on resolution, frame rate and video quality a camera using H.264 compression may generate the following bit rates:
- 352 x 240 @ 30 FPS, high quality = 0.4 Mbps
- 720 x 576 @ 30 FPS, high quality = 1 Mbps
- 1280 x 720 @ 30 FPS, high quality = 2 Mbps
- 1920 x 1080 @ 30 FPS, high quality = 4 Mbps
- 1920 x 1080 @ 30 FPS, medium quality = 2.8 Mbps
- 1920 x 1080 @ 30 FPS, low quality = 2 Mbps
- 1920 x 1080 @ 10 FPS, high quality = 2.4 Mbps
- 1920 x 1080 @ 10 FPS, low quality = 1.2 Mbps
Bitrate table (H.264 estimate) in Mbps with high quality image (x0.7 = standard quality):
|
Frames |
1 |
6 |
10 |
15 |
30 |
|---|---|---|---|---|---|
| CIF (352x240) | 0.16 | 0.2 | 0.24 | 0.3 | 0.4 |
| D1 0.4M (720x576) | 0.4 | 0.5 | 0.6 | 0.75 | 1 |
| 720p 1M | 0.8 | 1 | 1.2 | 1.5 | 2 |
| SXGA 1.3M (1280x1024) | 1 | 1.25 | 1.5 | 1.9 | 2.5 |
| HD 2M (1920x1080) | 1.6 | 2 | 2.4 | 3 | 4 |
| 3M | 2 | 2.5 | 3 | 3.75 | 5 |
| 5M | 3.2 | 4 | 4.8 | 6 | 8 |
Please note that these are estimates providing a high quality image under most conditions. If the scene is less complex (indoors with little detail and not much motion) or the camera has very little noise (daylight, good DNR) the bit rate can be lowered further. Avoid using less than half of the indicated values.
If video compression is set to lower quality or capped at a defined max bandwidth, the bit rate can be significantly lower at the cost of lower image quality. DNR can further reduce bandwidth, especially for grainy night images, but shows less detail during motion.
FortiRecorder maximum bandwidth
Recommended maximum total camera bandwidth usage for different FortiRecorder models and setup environments.
|
FRC Model |
Continuous |
Continuous with NAS |
Continuous with motion |
Continuous with motion and NAS |
|---|---|---|---|---|
|
FRC-100D |
90 Mbps |
TBC |
35 Mbps |
TBC |
|
FRC-200D gen1 |
90 Mbps |
55 Mbps |
50 Mbps |
50 Mbps |
|
FRC-200D gen2 |
135 Mbps |
135 Mbps |
130 Mbps |
130 Mbps |
|
FRC-400D |
170 Mbps |
160 Mbps |
140 Mbps |
130 Mbps |
|
These values have been determined experimentally in a lab setting and do not represent hard limits. Performance degrades gradually with symptoms like slow response or dropped video frames. Real world performance depends on many factors, including network environment and NAS types. The motion detection rate in the table above was 13% based on one detection of 40s length every 5 minutes per camera. |
Storage capacity
Video retention depends on the available storage capacity and the total amount of video bandwidth from the cameras.
To calculate storage capacity, note that a 1TB HD stores 1 camera configured to consume 1Mbps for approximately 100 days.
Video retention period in days for hard drive capacities:
|
|
FortiRecorder 100D with 1 TB HD |
FortiRecorder 200D with 3 TB HD |
FortiRecorder 400F with 4TB HD |
FortiRecorder 200D with 3 TB HD plus 16 TB remote storage |
FortiRecorder 400F with 32 TB HD |
|---|---|---|---|---|---|
|
1MP@30 FPS standard video quality = 1.4 Mbps |
72 |
218 |
291 |
1381 |
2327 |
|
2MP@15 FPS standard video quality = 2.1 Mbps |
48 |
145 |
194 |
921 |
1551 |
|
3MP@10 FPS high quality video = 3 Mbps |
34 |
102 |
136 |
645 |
1086 |
|
3MP@30 FPS high quality video = 5 Mbps |
20 |
61 |
81 |
387 |
651 |
For more information about bandwidth consumption calculation, see the FortiCamera Bandwidth Calculator User Guide on http://docs.fortinet.com/d/fortirecorder-forticamera-bandwidth-calculator-user-guide.
In practice, Fortinet suggests to use the numbers provided in the bandwidth calculator as a starting point and then adjust them after installation to achieve the desired balance between quality and bandwidth.
Client Performance
If you need to display 8 or more camera live views, you may need to configure the second camera stream so that viewing is done at a lower frame rate or resolution, depending on how powerful the client PC is. RAM is less important than CPU for rendering video.
Video playback is very CPU intensive. If you are experiencing choppy video playback and cameras “freezing” during playback, you likely have a client performance problem. Use the diagnostic tools available on your client OS and look at the CPU usage when you are experiencing video problems. If possible, keep the CPU usage below 50%.
To optimize client performance, use the video and camera profiles to define and assign a second video stream for each camera. To increase the number of live views the client computer can display, or to reduce the CPU requirement for a given number of live views, reduce the resolution, quality and/or frames per second of the second video streams.
Ten FPS is a good general setting for live views, which provides a reasonable frame rate for the live views, but significantly reduces the load on the client (compared to 30 FPS which is more ideal for higher traffic area surveillance).