NVR Storage: How Many Days Is Enough

Storage is where surveillance system design gets specific. Camera selection, NVR platform, and camera placement are all important decisions, but storage sizing is where the math happens. An under-sized storage system runs out of space and starts overwriting footage before the retention window the client expects. An over-sized system wastes money. Neither outcome is acceptable on a properly designed installation.

The storage question is tied directly to retention requirements, recording mode, camera resolution, and RAID configuration. Each of these factors has a direct impact on the number. This article works through each of them with real numbers so that the storage decision can be made correctly. Our surveillance practice includes storage sizing as a standard deliverable on every project proposal.

The Bitrate Math

Storage sizing begins with bitrate. A 4K camera recording at H.265 compression generates between 6 and 10 Mbps depending on scene complexity. A camera covering a static indoor hallway generates less than a camera covering a driveway with trees, varying light, and weather movement. A useful planning figure is 8 Mbps for a typical exterior 4K camera at H.265. which is the compression standard for all current professional surveillance systems. Using H.264 instead of H.265 roughly doubles the storage requirement for the same resolution and quality, which is why H.264 is no longer appropriate for new installations.

The daily storage calculation for one camera: 8 Mbps equals 1 MB per second, which is 3,600 MB per hour, which is 86,400 MB or approximately 86 GB per day at continuous 24-hour recording.

For a 16-camera estate system at 30 days of continuous retention: 86 GB x 16 cameras x 30 days = approximately 41 TB of raw footage data. For 60 days, that doubles to approximately 82 TB. For 90 days, approximately 123 TB.

These are raw data figures, not installed storage figures. The installed storage requirement is higher because of RAID overhead and because NVR systems do not fill drives to 100 percent before flagging storage limits. A practical installed storage target for 30-day continuous retention at 16 cameras is 48 to 56 TB in a RAID 5 or RAID 6 configuration. For 60-day retention, plan for 96 to 112 TB installed.

30 vs 60 vs 90 Day Retention: What Actually Matters

The retention window question is not purely a technical decision. It is a legal and insurance decision that has technical consequences.

Thirty days is the most common minimum in residential surveillance, and it is frequently not enough. The problem with 30 days is that incidents are often not discovered within that window. A slow theft from a property. a contractor removing materials over multiple weeks. may not be noticed until 35 or 40 days after it began. An insurance claim for a slip-and-fall on a front walkway may be filed 45 days after the event. If the footage has been overwritten, it does not exist.

Sixty days is a more defensible standard for luxury residential properties. It covers the large majority of insurance claim scenarios and most civil litigation discovery requests where the incident date is known. For properties where the client has specific legal or insurance requirements. a home used for business purposes, a property with staff and regular contractor access, a home in a community with HOA-level security requirements. 60 days should be the minimum.

Ninety days is the standard for commercial properties in regulated industries and for residential estates where the client has been advised by counsel to maintain longer retention. At 90 days continuous, a 16-camera 4K system requires approximately 140 to 160 TB of installed RAID storage. That is a meaningful infrastructure investment, but it is the correct answer when the requirement is genuine.

Motion-Only vs Continuous Recording

Motion-only recording reduces storage consumption dramatically. A well-placed exterior camera at a residential property with normal activity generates 3 to 6 hours of motion-triggered footage per day rather than 24 hours of continuous footage. That is a 75 to 87 percent reduction in storage consumption. which means a 30-day motion-only system requires roughly the same storage as a 4 to 6 day continuous system for the same camera count.

The risk is real. A camera configured for motion-only recording stores no footage of events that fail to trigger the motion detector. A camera positioned to cover a gate area with a consistent wind-blown tree in the background may have its motion sensitivity set low to reduce false alerts. and may miss a slow-moving approach by someone who understands how to move under the detection threshold. A camera that is in the direct path of a vehicle that deactivates the motion detector through its angle of approach may fail to trigger.

The professional standard is continuous recording as the baseline with AI smart detection layered on top. The system records everything. The AI flags which events involve persons, vehicles, packages, or other defined object types. The client reviews AI-flagged events rather than raw footage. The storage cost of continuous recording is justified because it eliminates the failure mode of missing events that motion-only recording creates.

RAID Redundancy for Surveillance NVRs

A surveillance NVR without RAID is not a professional installation. Hard drives in surveillance NVRs run at high utilization. continuous writes from multiple camera streams. and drive failure rates in always-on surveillance applications are meaningfully higher than in lightly used storage systems. A single drive failure in a non-RAID NVR means footage loss from the day of failure until the drive is replaced and the system is rebuilt.

RAID 5 is the minimum acceptable configuration. RAID 5 distributes parity data across all drives, allowing the system to continue operating and recovering after a single drive failure. The capacity overhead is approximately one drive’s worth of capacity across the array. a 4-drive RAID 5 array provides three drives of usable capacity. The practical installed-to-usable ratio is approximately 75 percent for RAID 5.

RAID 6 provides dual-drive failure tolerance. This is appropriate for large estates with long retention requirements where losing footage during a rebuild window is unacceptable. RAID 6 overhead is approximately two drives. a 6-drive RAID 6 array provides four drives of usable capacity, approximately 67 percent efficiency. For a 16-camera system with 60-day retention requirements, RAID 6 is the appropriate choice.

RAID redundancy protects against drive failure. It does not protect against NVR theft, fire, or flood. For high-value estates where the surveillance footage itself is a critical asset. properties with ongoing legal matters, properties with documented threats, properties that are unoccupied for extended periods. an offsite backup strategy is required. This may involve a secondary NVR at a separate location, cloud archiving of AI-tagged events, or both.

AI Event Tagging: The Storage Compressor

AI event tagging changes the storage question in an important way, though not by reducing the total footage stored.

What AI event tagging does is make specific footage instantly accessible. A 90-day retention window on a 16-camera system contains approximately 34,560 camera-hours of footage. Without AI tagging, finding a specific person or vehicle event in that archive requires either knowing the timestamp precisely or reviewing large amounts of footage manually. With AI tagging, the system has indexed every person, vehicle, and package event with a timestamp and classification. Finding all footage of a specific vehicle entering the property over the past 60 days is a seconds-long search rather than a multi-hour review.

The storage compression claim is more specific: AI event tagging allows a client to maintain shorter continuous recording retention while maintaining longer event-only retention. If the system stores continuous footage for 30 days but AI-tagged person and vehicle events for 90 days, the storage requirement is the 30-day continuous figure plus a relatively small increment for 60 additional days of event-only clips. Person and vehicle events at a typical residential property might represent 3 to 5 percent of total continuous footage by duration. Ninety days of event-only footage is a fraction of 90 days of continuous footage.

This is the hybrid retention model that Restrepo Innovations recommends for most luxury residential properties: 60-day continuous retention as the baseline, AI event tagging indexing the full archive, and 90-day event-clip retention for person and vehicle events specifically. It provides the legal defensibility of a long event archive while keeping the total storage requirement manageable. For the network infrastructure that supports this kind of storage architecture, see our post on estate home network infrastructure and our networking practice page.

“The question isn’t how many days of storage you have. It’s whether the footage you need still exists when you need it. Those are different questions.” __EMDASH_PROTECT_0__

If you are sizing a surveillance system for a luxury residence or estate property in New Jersey, Connecticut, or New York, our team can run the storage calculation for your specific camera count, resolution, and retention requirement. Contact us at 201.405.2022 or schedule a consultation at Calendly.