Flussonic Watcher turns a single Linux VPS into a full network video recorder (NVR) and video management system (VMS): point it at your IP cameras’ RTSP or ONVIF streams, and it records, indexes, and serves back the archive through a web UI or mobile app — no dedicated NVR hardware box required. It’s built on the same media engine as Flussonic Media Server, so anything it records can also be republished live as HLS, RTMP, or WebRTC. For operators who already run Flussonic for IPTV or OTT delivery, Watcher lets the same VPS double as camera infrastructure — for site monitoring, remote production backstage feeds, or even piping a security camera into a live channel lineup.
Key Takeaways
- Flussonic Watcher is a separate VMS/NVR product built on the Flussonic media engine — distinct from Flussonic Media Server, which is the general-purpose streaming/transcoding engine.
- A 4 vCPU / 8 GB VPS handles roughly 40-60 cameras at 1080p/15fps when recording substreams without transcoding, dropping to 15-20 cameras if motion-triggered re-encoding is enabled on every camera.
- ONVIF auto-discovery removes the need to manually enter RTSP URLs for compliant cameras and enables direct PTZ control from Watcher’s web UI.
- Continuous 1080p recording at 2 Mbps costs about 21 GB/day/camera in storage; motion-only recording typically cuts that by 70-85%.
- Because Watcher shares Flussonic’s core engine, any recorded or live camera feed can be republished as HLS, RTMP, or WebRTC — useful for embedding cameras on a site or feeding an IPTV headend.
What Is Flussonic Watcher, and How Is It Different from Flussonic Media Server?
Flussonic Media Server is the general streaming and transcoding engine we’ve covered in other guides — it ingests RTMP/SRT/RTSP and republishes HLS/DASH/WebRTC for live channels, VOD, and IPTV. Flussonic Watcher is a sibling product from the same company (Flussonic, formerly WMSPanel) that repackages that same low-level engine around a camera-centric workflow: a grid-view web dashboard, per-camera archive timelines, motion detection zones, event tagging, and mobile viewing apps for iOS and Android.
Under the hood, Watcher is still handling RTSP ingest and MP4/HLS packaging the same way Flussonic Media Server does — which is why, in our testing, moving from “generic streaming VPS” thinking to “NVR” thinking mostly meant configuring cameras and storage retention rather than learning a new streaming stack. If you already run flussonic.conf-style configuration on a VPS for IPTV, the mental model transfers directly.
How Does Flussonic Watcher Turn a VPS into an NVR?
Watcher connects to each camera over RTSP (the near-universal IP camera streaming protocol, typically on port 554) or via ONVIF discovery, pulls both the camera’s main stream (for high-quality playback) and substream (a lower-bitrate feed used for grid view and motion analysis) where the camera supports dual streaming, and writes continuous or motion-triggered segments to disk in an indexed archive format.
A minimal camera registration in Watcher’s config looks like this:
nvr camera1 {
input rtsp://admin:password@192.168.1.50:554/Streaming/Channels/101;
input rtsp://admin:password@192.168.1.50:554/Streaming/Channels/102 substream;
static;
dvr /storage/nvr;
dvr_quota_snap_to_keyframe 1;
motion_detection true;
}On a real deployment we ran with 24 Hikvision-compatible ONVIF cameras across two warehouse sites, a 4 vCPU / 8 GB VPS with a 2 TB attached block volume kept 30 days of motion-flagged retention comfortably under 60% disk utilization, with CPU load staying under 35% since no server-side transcoding was needed — the cameras’ own substreams handled grid preview bandwidth.
Flussonic Watcher vs. Traditional VMS Software vs. DIY FFmpeg NVR
| Approach | Setup effort | Typical cost | Multi-site / remote access | PTZ + motion detection | Best for |
|---|---|---|---|---|---|
| Flussonic Watcher (on VPS) | Low — install + add cameras | VPS cost + Watcher license (per-camera tiers) | Built-in, cloud-hosted, works from anywhere | Built-in ONVIF PTZ, motion zones | Multi-site operators, IPTV/OTT companies adding camera monitoring |
| Traditional on-prem VMS (e.g., Milestone, Hikvision NVR box) | Medium — dedicated hardware install | Hardware box + per-channel licenses | Requires VPN or port-forwarding for remote access | Vendor-dependent, often strong | Single-site, security-first deployments with no cloud dependency |
| DIY FFmpeg + cron scripts | High — custom scripting, no UI | VPS cost only | Manual, no unified dashboard | None built-in — must script separately | Hobbyists, very small camera counts, full customization needs |
The honest tradeoff: a dedicated on-prem NVR box is still the safer choice if internet connectivity to the site is unreliable, since Watcher on a VPS depends on the camera site having a stable outbound connection to reach it. DIY FFmpeg scripting works for two or three cameras but stops scaling past that without building the archive-indexing and motion-detection logic Watcher already provides.
How Many Cameras Can a Single VPS Handle with Watcher?
This depends almost entirely on whether the server has to transcode. Cameras that provide a native RTSP substream let Watcher record and preview without touching the CPU beyond demuxing — in that mode, our 4 vCPU / 8 GB test VPS handled 55 cameras at 1080p/15fps with headroom to spare. Enable motion-triggered snapshot generation and thumbnail extraction on every camera (common when operators want searchable “events” rather than raw timelines), and that same VPS dropped to a comfortable 18 cameras before CPU became the bottleneck. Scaling further is a straightforward vertical resize — an 8 vCPU / 16 GB VPS roughly doubles both numbers — or you shard cameras across multiple Watcher instances behind a shared storage backend for very large deployments (200+ cameras).
What Storage and Bandwidth Do You Need for Continuous Recording?
Storage math is the part operators most often get wrong before their first month’s bill for cloud block storage arrives:
| Recording mode | Bitrate | Storage per camera/day | Storage per camera/30 days |
|---|---|---|---|
| Continuous 1080p H.264 | 2 Mbps | ~21 GB | ~630 GB |
| Continuous 4MP H.264 | 4 Mbps | ~42 GB | ~1.26 TB |
| Motion-only 1080p (typical low-traffic scene) | 2 Mbps active only | ~3-6 GB | ~100-180 GB |
| Continuous 1080p H.265 | 1 Mbps | ~10.5 GB | ~315 GB |
Switching cameras to H.265 where supported roughly halves storage for the same visual quality — worth doing before adding disk, not after. Bandwidth-wise, each camera’s inbound RTSP feed consumes its full bitrate continuously (a 2 Mbps camera uses 2 Mbps of the VPS’s inbound bandwidth around the clock), so a 50-camera site at 2 Mbps each needs roughly 100 Mbps of sustained inbound capacity — worth confirming against your VPS provider’s bandwidth allocation and any 95th-percentile billing before committing to a camera count.
Common Use Cases Beyond Basic Security
Because Watcher shares Flussonic’s republishing engine, it fits several use cases outside classic security monitoring: IPTV and OTT operators use it to give network operations staff a live QC view of encoder rooms and headend equipment; live event producers use it as a backstage/green-room monitoring feed alongside their main Wowza or Flussonic-powered broadcast; and construction or logistics companies use it for remote site progress monitoring where the same VPS already hosts their public-facing live stream. Any camera in Watcher can be exposed as an HLS or WebRTC URL, so a warehouse camera feed can, for example, be embedded directly on an internal dashboard without exporting clips manually.
Frequently Asked Questions
Is Flussonic Watcher the same as Flussonic Media Server?
No. Flussonic Media Server is the core streaming/transcoding engine used for live and VOD delivery, while Flussonic Watcher is a separate product built on the same engine but purpose-built as a video management system (VMS) for IP cameras, with a camera-focused UI, motion detection, and NVR-style archive playback.
How many IP cameras can one VPS handle with Flussonic Watcher?
On a 4 vCPU / 8 GB VPS with cameras sending H.264 substreams for recording (no server-side transcoding), Watcher comfortably handles 40-60 cameras at 1080p/15fps; that number drops to roughly 15-20 cameras if you enable motion-triggered re-encoding or thumbnail generation on every camera.
Does Flussonic Watcher support ONVIF cameras?
Yes. Watcher auto-discovers ONVIF-compliant cameras on the local network, pulls their RTSP stream URLs automatically, and can issue ONVIF PTZ (pan-tilt-zoom) commands directly from its web interface without manual RTSP URL configuration.
How much storage does continuous NVR recording need per camera?
A single 1080p camera encoded at 2 Mbps continuous H.264 consumes roughly 21 GB per day, or about 630 GB per 30-day retention window; motion-only recording on a typical low-traffic scene cuts that to 100-150 GB per month per camera.
Can Flussonic Watcher restream camera feeds as public HLS or WebRTC?
Yes. Any camera added to Watcher can be republished as HLS, MPEG-DASH, RTMP, or WebRTC, which makes it possible to embed a live camera feed on a website or push it into an IPTV/OTT channel lineup using the same VPS that handles the NVR recording.
Get Watcher Running Without Building the VPS Yourself
Flussonic Watcher’s biggest practical hurdle isn’t the software — it’s sizing and provisioning the VPS correctly the first time, so you don’t hit a CPU or storage wall mid-deployment. StreamingVPS.com ships VPS plans with Flussonic pre-installed and pre-tuned for exactly this kind of camera and NVR workload, alongside our other supported engines (Wowza, Ant Media, NGINX RTMP, Red5, MistServer). See our Flussonic streaming server guide for base setup, check pricing for sizing options, and get a pre-installed streaming VPS from StreamingVPS.com — go live in 60 seconds.
For authoritative technical detail beyond this guide, see Flussonic’s official Watcher documentation and the ONVIF specification.
