5 Video surveillance storage at city or campus scale is a different problem from storing footage for a single building. A city, transport authority, university, hospital campus or correctional system may run thousands of cameras, many recording around the clock, across dozens of sites. The storage behind them has to absorb a constant stream of writes, keep footage for defined retention periods, survive hardware failures without losing evidence and make the right clip available quickly when an incident occurs. This article explains what large-scale video surveillance storage has to deliver, the architectures organizations use and how to evaluate options as camera counts and resolutions grow. It is written for storage engineers and security systems architects. For sizing math, see how to calculate CCTV storage requirements. What makes surveillance storage different Surveillance workloads have an unusual profile: Write-heavy and continuous. Cameras write 24 hours a day. Most footage is never viewed. Sequential and predictable. Each camera produces a steady stream at a known bitrate. Large and growing. Higher resolutions, more cameras and longer retention all increase capacity. Time-bounded retention. Most footage is deleted automatically after a set period, while footage tied to incidents is preserved longer. Evidence-grade when it matters. Clips used in investigations or court must be intact and their handling documented. Distributed. Cameras are spread across many locations with varying network connectivity. Storage that is excellent for databases or virtual machines may be poorly matched to this profile, and vice versa. The role of the video management system A video management system (VMS) records camera streams, manages retention, provides live and recorded viewing and exports evidence. The VMS usually decides how footage is written and organized. Many VMS platforms use tiers: Live or recording tier: where streams are first written, often on recording servers with local or attached storage. Archive tier: where footage is moved after a period, often to larger, cheaper storage. Evidence tier: where exported or protected clips are kept for investigations. The storage team’s design must match the VMS’s capabilities. Many current VMS platforms can archive to network storage, and some support S3-compatible object storage. Storage architectures for large deployments NVRs and edge recorders Network video recorders at each site combine recording software and local disks. They are simple and work without central infrastructure, but at scale they create hundreds of storage islands, each with its own failure risk, capacity limit and maintenance burden. Recording servers with SAN or NAS Larger deployments run VMS recording servers connected to shared SAN or NAS storage. This centralizes capacity and improves resilience, but scale-up arrays can become expensive and require migrations when replaced. Scale-out storage Scale-out file or object storage pools capacity across many nodes and grows by adding nodes. It suits large surveillance estates because capacity and throughput grow together and hardware can be refreshed without moving footage. Object storage also supports immutability for evidence protection. Hybrid edge and core Many cities and campuses combine edge recording for resilience with central archiving for retention and investigation. Edge devices keep recent footage locally in case of network outages, and footage is moved or replicated to central storage for longer retention. What to evaluate Sustained write performance The storage must handle the combined bitrate of all cameras continuously, with headroom for peaks and failures. Rebuilds after a drive failure must not cause dropped frames. Test write performance during degraded states, not just when everything is healthy. The scality.com blog’s article on object storage rebuilds explains what happens during failures. Capacity and growth Capacity is driven by camera count, resolution, codec, frame rate, recording mode and retention. Upgrading cameras to higher resolution can multiply storage needs. Plan for growth and choose storage that expands in small steps. Retention management The VMS typically deletes footage when retention expires. Storage should handle large-scale deletion efficiently and support longer retention for specific footage under legal hold. Evidence integrity Footage used as evidence must be preserved intact, with access and export logged. Immutable storage for exported evidence or protected recordings prevents deletion or alteration. The principles overlap heavily with digital evidence chain of custody. Resilience Losing footage at the moment it is needed is the worst outcome. Look for protection against drive and node failure, options for replication to another site and recording continuity during maintenance. Cybersecurity Cameras and VMS platforms are frequent targets. Storage should be isolated from general IT networks where possible, use separate credentials and support encryption. Ransomware that encrypts surveillance archives can destroy evidence and disrupt safety operations. Privacy and compliance Surveillance footage is personal data under GDPR and similar laws. Storage design must support retention limits, access control, audit logging and secure deletion. Cost per camera over time Compare the cost per camera per year, including storage, servers, facilities and refresh. Centralized scale-out storage often lowers this at large scale by improving utilization and avoiding per-site hardware. Network design between sites and storage At city or campus scale, the network often determines what storage design is possible. Cameras on a well-connected campus can stream directly to central recording servers. Cameras on poles across a city, at remote transit stations or on buses may rely on wireless or cellular links with limited and variable bandwidth. Common approaches include: Recording at the edge and sending footage to central storage on a schedule or on demand. Dual streams, with a lower-resolution stream sent centrally for live viewing and a full-resolution stream kept locally. Regional aggregation points that collect footage from nearby cameras before forwarding it to the core. Dedicated surveillance networks, segmented from corporate IT for both performance and security. Whatever the approach, size network links for continuous recording plus bursts when investigators retrieve large amounts of footage. A storage platform that can sit at both regional and central sites, using the same management and protection, simplifies operations across this kind of topology. Operations at scale With thousands of cameras, small failures are constant. Disks fail, cameras go offline and links drop. Monitoring should cover recording gaps per camera, storage capacity and health, and archive job success. Automated alerts for cameras that stop recording are as important as storage alerts, since a camera that silently stopped writing weeks ago is only discovered when footage is needed. Multi-agency and shared platforms Cities often share surveillance infrastructure across departments: police, transport, traffic management and public works. Campuses may share across security, facilities and operations. Shared platforms reduce cost and simplify investigation, but require clear access rules, separate retention per department where needed and audit trails showing who viewed or exported footage. Storage that supports multi-tenancy, with separate accounts and policies per department, makes this easier. Planning for higher resolutions and analytics Camera technology keeps moving toward higher resolutions and more on-camera or server-side video analytics. Analytics can reduce storage, for example by recording at full quality only when motion or events are detected, or increase it, by keeping metadata and higher-quality footage for analysis. Modern codecs such as H.265 reduce bitrate compared with older codecs. The net effect varies, so revisit the storage plan whenever camera or analytics policies change. Checklist: video surveillance storage at scale Inventory cameras by resolution, codec, frame rate and recording mode. Calculate aggregate bitrate and capacity for each retention tier. Confirm the VMS’s supported storage types, including S3. Test sustained writes during drive and node failures. Choose storage that scales in small increments without migration. Protect evidence with immutability and logged access. Isolate storage networks and credentials from general IT. Design replication or edge buffering for resilience. Apply retention limits and secure deletion for privacy compliance. Model cost per camera per year over five years. Putting it together Video surveillance storage at city or campus scale needs to absorb continuous writes from thousands of cameras, keep footage for defined periods, preserve evidence intact and remain affordable as resolutions and camera counts grow. Centralized scale-out storage, often object storage, paired with a capable VMS and edge buffering where networks are unreliable, meets those needs better than hundreds of separate recorders. Start from the camera inventory and retention policy, test write performance under failure and choose a platform that grows without migrations. Frequently asked questions What storage is best for large video surveillance systems? Centralized scale-out storage is common at city or campus scale, often combined with edge recording for resilience. It pools capacity, scales by adding nodes and simplifies management compared with many NVRs. How much storage does a city surveillance system need? It depends on camera count, resolution, codec, frame rate, recording mode and retention. Large deployments often reach hundreds of terabytes to petabytes. Can video surveillance use object storage? Yes. Many VMS platforms can archive to network storage, and some support S3-compatible object storage directly. Check your VMS’s supported options. How do you protect surveillance footage used as evidence? Preserve it in immutable storage, log every access and export and verify its integrity when it is produced. Is surveillance footage subject to GDPR? Yes, footage of identifiable people is personal data under GDPR, so retention limits, access control and secure deletion apply. Further reading CCTV storage calculation VMS archive to S3 object storage CCTV footage retention NVRs vs centralized storage Body-worn camera storage
