6 Many organizations running large video surveillance estates ask whether their video management system (VMS) can archive footage to S3 object storage. In many cases the answer is yes. A growing number of VMS platforms support S3-compatible storage as an archive target, either directly or through a gateway, and object storage suits surveillance archives well: it scales by adding nodes, handles very large capacities, protects data with erasure coding and supports immutability for evidence. The details, however, depend on the VMS, its version and how it writes and reads footage. This article explains how VMS archiving to S3 object storage typically works, the integration models, what to evaluate and how to test before committing. For the broader picture, see our hub on video surveillance storage. Why archive surveillance footage to object storage Surveillance archives are large, write-heavy and mostly untouched. Footage is retained for a set period and then deleted, except for clips preserved as evidence. That profile matches object storage: Scale: archives can reach petabytes as camera counts and resolutions grow. Cost: object storage on standard servers usually costs less per terabyte than high-performance arrays. Durability: erasure coding protects footage across drives and nodes. Immutability: object lock can prevent evidence from being deleted or altered. Simplicity: one pool serves many recording servers and sites, instead of separate arrays per server. Refresh without migration: nodes can be replaced while the archive stays online. How VMS platforms use storage Most enterprise VMS platforms use a tiered approach: Recording storage: where live streams are written first, usually fast local or attached storage on recording servers. Archive storage: where footage is moved after a period, often hours or days, for the rest of its retention. Evidence or export storage: where protected clips are kept for investigations. S3 object storage is most commonly used for the archive tier. Some platforms can also record directly to object storage, though recording tiers typically stay on local storage to absorb continuous writes with predictable latency. Integration models Native S3 archiving The VMS writes archived footage directly to an S3 bucket using the S3 API. It keeps its own database of what is stored where and retrieves footage when users play back archived video. This is the cleanest model when supported, since it avoids extra components. Cloud or object archive modules Some VMS platforms offer separate archiving or cloud modules that move footage to object storage on a schedule. These may support public cloud, S3-compatible on-premises storage or both. Check whether on-premises S3-compatible targets are supported, not only specific public clouds. File gateway If the VMS supports only file protocols, a gateway can present object storage as an SMB or NFS share. This works for some archive scenarios but adds a component to size and manage, and performance depends on the gateway’s cache. Storage-side tiering Some storage platforms tier older files from a file system to object storage transparently. The VMS writes to a file share as usual, and the storage layer moves older data. This can work, but playback of tiered footage may be slower, and the VMS may not be aware of where data actually lives. Always confirm the supported model, versions and any certification for the specific VMS and storage combination. What to evaluate Write throughput Archiving moves large volumes of footage from recording servers to object storage, often continuously. The object store and network must sustain that rate alongside other workloads. Calculate archive throughput from total camera bitrate, as described in how to calculate CCTV storage requirements. Playback performance Investigators expect archived footage to play back quickly. Test playback and scrubbing through archived footage, as well as export of long clips. On-premises object storage on the same network as the VMS usually delivers good performance; public cloud archives can add latency and retrieval or egress costs. Object size and count How the VMS packages footage into objects affects efficiency. Large segment files are efficient; very small objects increase request rates and metadata load. Ask the vendor how footage is packaged and test at realistic scale. Retention and deletion The VMS usually deletes footage when retention expires. Confirm that deletions on object storage are handled efficiently and that capacity is reclaimed. Some deployments use bucket lifecycle rules as a backstop. Immutability and evidence For evidence, check whether the VMS supports writing protected footage with object lock, or whether a separate immutable bucket is used for exports. Object lock prevents deletion or alteration during the retention period, supporting chain of custody. Resilience If object storage becomes unavailable, recording should continue to local storage, with archiving catching up later. Test that behavior. For the archive itself, consider replication to another site or a platform that spans sites. Security Surveillance storage should be isolated from general IT where possible, with dedicated credentials for the VMS and encryption in transit and at rest. Multi-site deployments Organizations with many sites may archive locally at each site, centrally at a core data center, or both. Object storage can be deployed at regional and central sites with replication between them, giving local playback and central retention. Sizing the object store for archiving Size the archive from camera bitrates and archive retention. If recording servers keep, for example, three days locally and footage is retained for 30 days in total, the object store holds 27 days of footage for every camera, plus protection overhead, preserved evidence and headroom. Throughput must match the aggregate camera bitrate on average, with extra capacity so the archive can catch up after a maintenance window or outage. Plan growth for new cameras and resolution upgrades so expansion happens in small, scheduled steps. Public cloud or on-premises S3 Both can work. Public cloud object storage avoids running infrastructure, but surveillance archives are large and constantly written, so storage costs grow quickly, and playback or export can incur retrieval and data transfer charges. Bandwidth from many sites to the cloud can also be a constraint. On-premises S3-compatible object storage keeps footage local, avoids transfer fees and provides predictable costs at large scale. Many public sector organizations also prefer to keep footage under direct control for privacy and data protection reasons. How to test before committing A pilot should prove the integration in your environment: Confirm supported versions and configuration for your VMS and storage. Archive footage from a representative set of cameras for at least a week. Measure archive throughput and confirm it keeps up with recording. Test playback, scrubbing and export of archived footage. Simulate object storage unavailability and confirm recording continues. Simulate a storage node failure and confirm playback and archiving continue. Test retention-based deletion and capacity reclamation. Test evidence protection with object lock or immutable buckets. Migration from existing archives Moving existing footage to object storage is often unnecessary because surveillance retention is short. A common approach is to point new archiving at object storage and let old archives expire naturally on existing storage. Longer-retained evidence can be migrated selectively. This avoids a large migration project. Checklist: VMS archive to S3 object storage Confirm S3 support model, versions and certification with the VMS vendor. Keep recording on local storage and use object storage for archive. Size object storage for capacity and continuous archive throughput. Test playback and export performance from the archive tier. Check object packaging, request rates and deletion efficiency. Use object lock or immutable buckets for evidence. Ensure recording continues if the archive is unavailable. Isolate storage networks and credentials. Plan multi-site archiving and replication. Let old archives expire rather than migrating them. Putting it together A VMS can archive to S3 object storage when the platform supports it, and for large surveillance estates it is often the most scalable and economical archive tier. Keep recording on local storage, archive to object storage sized for continuous throughput and playback, protect evidence with immutability and test failure behavior before going live. With that in place, the archive can grow with the camera estate without repeated hardware migrations. Frequently asked questions Can video management systems use S3 storage? Many VMS platforms support S3-compatible storage for archiving, natively or through modules or gateways. Check your VMS’s supported options and versions. Should cameras record directly to object storage? Usually recording stays on local storage on recording servers, with footage moved to object storage for the archive tier. Is playback from object storage slow? On-premises object storage on the same network as the VMS typically plays back well. Test playback and export during a pilot. Can object storage protect surveillance evidence? Yes. Object lock can make preserved footage immutable for a defined period, preventing deletion or alteration. Do I need to migrate old footage to object storage? Often not. Because surveillance retention is short, many organizations let old archives expire and send new footage to object storage. Further reading Video surveillance storage CCTV storage calculation CCTV footage retention NVRs vs centralized storage Digital evidence storage
