Hybrid Video Surveillance Architecture: What Stays Local and What Moves to the Cloud

A surveillance system must keep recording while operators view live feeds and review footage. When that work spans multiple sites, system integrators need to decide which functions each site should handle and which services should be shared centrally or through the cloud.

Interest in hybrid deployments is growing. The 2026 Genetec State of Physical Security Report, based on responses from 7,368 physical security professionals worldwide, found that 72% of consultants plan to specify hybrid deployments over the next five years. The report also says 71% of channel partners expect an increase in new cloud systems in 2026.

NEXCOM Smart Vision (GreenBase Technology Co., Ltd.) sees hybrid surveillance as an extension of local video infrastructure. The local NVR remains the operational anchor for high-volume recording and concurrent tasks such as live decoding and playback. Central or cloud services can extend access and coordination across locations. The VMS or VSaaS software and system integration determine which services are available; NEXCOM NViS 6712 provides the local hardware foundation.

The useful starting point is the work the system must perform. A workload’s data volume, continuity needs, access pattern, and software requirements help determine where it belongs.

Start With the Workload

At a local NVR, incoming video can serve two jobs at once. One path sends footage to storage for retention. Another decodes video for live viewing on a control-room display. Operators can monitor multiple feeds, open a selected camera for a closer look, or use playback while recording continues.

This matters when planning a system. A deployment sized for recording has different demands from one that also handles live viewing, playback, VMS services, database access, metadata management, or other applications.

To understand those demands, integrators need the camera count, bitrate, resolution, frame rate, codec, recording policy, and retention period. Together, these details inform both the storage estimate and the system’s capacity for concurrent tasks. Camera count alone cannot describe the full system demand.

Match Each Workload to a Starting Point

These are common starting points for hybrid surveillance design. Site requirements, connectivity, security policies, and software capabilities can shift the workload placement.

Use the table to map each function to a local, central, or cloud role. With those roles defined, planners can estimate the local storage needed for high-volume recording.

Keep High-Volume Video Close to Its Source

Continuous recording creates a large, persistent data stream. For a first-pass storage estimate, start with aggregate camera bitrate, recording time, and retention. Then validate the estimate against the full set of system demands described above.

Event-based recording can require less storage than continuous 24/7 recording. Extending retention from a few days to several months multiplies the requirement across every camera. For deployments with many cameras or long retention periods, keeping high-volume footage close to its source can make practical sense.

Storage costs add another consideration. Omdia’s Video Surveillance & Analytics 2026 research, covered in SecurityWorldMarket’s report on three strategic shifts shaping video surveillance, reports that the global video surveillance market grew 4.3% to $27 billion in 2025. Omdia forecasts 12.8% growth in 2026, driven primarily by component price escalation rather than higher sales volume. This makes storage planning a cost decision as well as a capacity decision. Omdia recommends reassessing storage infrastructure and exploring hybrid-cloud architectures.

Extend Access and Coordination Across Sites

Central or cloud services can help authorized staff review footage remotely, receive alerts, access metadata, or manage multiple locations. The available functions depend on the selected video management system (VMS) or Video Surveillance as a Service (VSaaS) solution. Raw video, selected clips, metadata, alerts, and management traffic each create different bandwidth and access requirements.

Consider a city security operation receiving an incident report from one of its public spaces. Local storage preserves the footage according to the site’s retention policy. The city’s access process determines who can retrieve the relevant video and records why it was requested. Central services may support remote review, alerts, or metadata management across locations, depending on the selected software package.

The placement decision follows the workflow: footage stays at the site where it is recorded, while authorized users can access relevant information across the organization. The hardware provides local video resources; the software and integration define how users retrieve and manage that information.

Place Analytics Where the Application Fits

Video analytics can run inside an intelligent camera, on a local video server, on a dedicated processing node, or in central infrastructure. The right location depends on what the application needs and where its data can be processed effectively.

SDM’s 2026 State of the Market: Video Surveillance describes video surveillance expanding into broader security workflows, operational efficiency, and business intelligence. As organizations apply more functions to video, system planners need to account for the added processing requirements.

Some deployments use a dedicated node for AI processing. Others may run selected applications alongside recording and viewing. Integrators should consider latency, software support, compute, bandwidth, power, and operating cost when deciding where processing belongs. The planned configuration should also be validated against resolution, frame rate, bitrate, decoding, and the tasks expected to run at the same time.

Turn the Workload Placement Plan into a Sizing Plan

Once the local and central roles are clear, integrators can work through five practical questions:

1. How much video does each site generate?

A continuous 1 Mbps stream generates approximately 10.8 GB per day using decimal storage units, before accounting for bitrate variation and implementation overhead. Multiply that estimate by stream count and retention period to establish a baseline.

2. What must keep working if connectivity is interrupted?

If a site loses its connection to central or cloud services, local recording should continue so no footage is lost. Site-level live monitoring and playback should also remain available where operators need them for immediate response or investigation. Functions that depend on cross-site access—such as remote review, centralized administration, or cloud-based services—can resume once connectivity is restored.

3. What information needs to cross the site boundary?

Specify whether central or cloud services need raw video, selected clips, metadata, alerts, or management traffic. Tie each flow to a user or operational process.

4. What protection does the footage require?

For EU deployments, surveillance footage that identifies individuals is subject to the GDPR, including requirements around purpose limitation, data minimization, retention, and security. In practice, footage should be kept only as long as necessary. The EDPB’s video surveillance guidelines note that recordings should generally be deleted after a few days, with retention beyond 72 hours requiring stronger justification; individual member states may also impose additional rules. These requirements should be considered alongside the footage’s evidentiary and operational value when defining retention and storage protection policies.

5. How might system demands change over time?

Account for the different replacement cycles of cameras, servers, storage, software, and networks so that upgrades can follow actual site needs.

Together, these questions help teams match local capacity to the required functions while giving central services a defined role.

Size Retention and Protection Separately

Retention and protection answer different planning questions. Retention sizing determines how much usable storage is needed for the recording schedule and retention period. Protection planning determines how the system should tolerate drive failure and what additional recovery measures the deployment requires.

A capacity-oriented JBOD configuration can prioritize usable storage and flexible configuration. A selected hardware RAID scheme can provide redundancy while reducing usable capacity relative to the raw storage installed. Teams should calculate the capacity required for retention first, then determine how much additional raw capacity the selected protection scheme requires. RAID can address selected drive-failure risks; it does not replace a separate backup or recovery plan where one is required.

Once retention and protection requirements are clear, planners can size the local node for both storage and concurrent video tasks.

NViS 6712 Provides Local Storage and Compute

To support those local requirements, NEXCOM NViS 6712 combines 12 × 3.5-inch HDD bays supporting drives up to 26TB each—up to 312TB of JBOD raw capacity—with two M.2 2280 NVMe slots and one M.2 SATA slot. Actual retention depends on the recording profile and selected protection scheme.

For concurrent video tasks, NViS 6712 supports Intel® Core™ Series 2 processors with up to 12 P-cores, 24 threads, 5.9 GHz Turbo, and 36MB L3 cache, plus up to 64GB DDR5 memory. Dual 2.5GbE interfaces and PCIe and M.2 expansion provide network connectivity and options for additional deployment requirements, including local processing applications.

Explore the NViS 6712 product news for more on its storage, compute, workload separation, and AI expansion capabilities.

Build a Hybrid Video Surveillance Architecture Around Operational Needs

Hybrid deployments are gaining ground as video supports broader security and operational needs. NEXCOM Smart Vision’s view is to keep high-volume recording and essential site operations local, extending access and coordination across sites as required. NViS 6712 provides the local storage and compute foundation, while the selected VMS or VSaaS shapes how it fits into the wider system. For system integrators, the next step is to assign each function to the place that best supports the people who rely on it.