Lead Forensics

When does a CCTV project need a storage area network rather than local storage?

A CCTV project needs a storage area network (SAN) when recording, retention or resilience requirements outgrow what individual servers with local disks can deliver. Typical triggers are very large camera counts, long retention periods, storage that must be shared across several recording servers, and a need for failover without losing recorded video. Smaller or self-contained sites are usually better served by local storage. The sections below cover the signals, the numbers, the exceptions and the checks that separate the two.

Which project signals point towards a SAN?

The strongest signals are many recording servers needing a shared pool of capacity, retention that exceeds what a single chassis can hold, and a requirement for recording servers to fail over to one another. When two or more of these apply, SAN storage is worth evaluating seriously.

In practice, these signals tend to appear together on enterprise projects:

  • Stranded capacity: with local storage, each recording server has its own disks. Some fill up while others sit half empty, and you cannot easily rebalance without moving cameras. A SAN presents one pool that is allocated where it is needed.
  • Failover recording: if a recording server fails, a standby server can only resume writing to the same archive if the storage is reachable by both. Storage that lives inside the failed server is not.
  • Growth that is hard to predict: campus, transport and city projects often add cameras in phases, change resolutions or enable analytics later. Storage that scales independently of compute is easier to expand than adding another full server each time.
  • Centralised data management: where a security or IT team wants one place to monitor, maintain and protect video storage, consolidation reduces the number of separate systems to look after.

Large multi-site deployments such as education campuses and public space networks often show these characteristics, which is why shared storage appears regularly in their designs. A real example is Secure Logiq’s Qatar University SAN storage case study.

How do bandwidth and retention drive the decision?

Bandwidth sets the sustained write rate your storage must handle, and retention multiplies that rate into total capacity. As aggregate bitrate and retention days rise, a single server’s disk bays and controller become the limiting factor, and pooled storage across several servers becomes more practical.

The basic calculation is straightforward. Total recording bandwidth, multiplied by the number of seconds in the retention period, gives the raw video volume before overheads. Because this depends on bitrate, compression, scene activity and recording mode, it should always be modelled with real project assumptions rather than rules of thumb. As a purely hypothetical illustration, a system recording at an aggregate 1 Gbps would write roughly 125 MB per second continuously, which is about 10.8 TB per day before RAID overhead. Thirty days of retention would then call for hundreds of terabytes of raw video storage. Figures like this are illustrative, not a performance guarantee for any product.

Two separate questions follow from this:

  • Can the storage sustain the write load? Surveillance involves continuous, largely sequential writing, with reads for playback and export on top. Sustained write performance matters more than peak benchmark figures.
  • Can it hold the retention period? Capacity must account for RAID parity, hot spares, free space headroom and future camera additions.

Analytics add another dimension. Metadata from AI analytics can create heavy, mixed workloads on the same disks as video, which affects archive stability and drive wear. Tools that simulate the camera load before deployment help here. Secure Logiq’s Logiqal Benchmark utility, for example, is designed to validate server and storage throughput limits ahead of physical installation, and the selecting a server guide covers the underlying sizing logic.

When is local storage still the better choice?

Local storage remains the better choice when a site has a modest camera count, a single recording server or independent servers per zone, and no need for shared capacity or cross-server failover. It is simpler, cheaper to deploy and has fewer components that can fail.

A SAN is not inherently the right answer for every CCTV deployment. It adds a storage fabric, switching, cabling, management and often additional licensing and skills requirements. Those costs are justified only when the benefits above are genuinely needed.

Local storage tends to be the sensible option when:

  • the project is a single site with predictable camera numbers and retention
  • each recording server can be sized to hold its own cameras’ video comfortably with headroom
  • the VMS handles redundancy through other means, such as camera-side edge recording or mirrored archives
  • the client’s team has limited experience with storage networking and no wish to take it on

High-density purpose-built servers can hold a large amount of video internally, so it is worth confirming that a single well-specified server does not already meet the requirement before adding a network layer. Our HD surveillance servers are designed for continuous sequential recording and often cover projects that might otherwise be assumed to need a SAN.

How do SAN and local storage compare on resilience?

A SAN can offer greater resilience for the system as a whole because storage survives the loss of a recording server, but it also introduces new shared points of failure. Local storage is simpler and isolates faults to one server, though a failed server takes its archive offline until repaired.

The trade-offs look like this:

  • Server failure: with a SAN, a standby recording server can potentially take over using the same storage, subject to VMS support. With local storage, the archive on that server is unavailable until it is recovered.
  • Disk failure: both architectures rely on RAID and hot spares to absorb drive failures. Neither removes the need for monitoring and prompt replacement.
  • Shared dependencies: a SAN adds controllers, switches and links. Without redundant paths and power, these can become single points of failure affecting every server that depends on them.
  • Fault isolation: a fault in local storage affects only the cameras on that server. A fault in shared storage can affect many.

It is also worth being clear about terminology. RAID protects against drive failure, not against deletion, corruption or site loss, so it is not a backup. Redundancy does not automatically guarantee uninterrupted recording either, because the VMS, network and power all have to cooperate. Resilience is a property of the complete architecture, not of the storage type alone.

What should be checked before specifying a SAN?

Before specifying a SAN, confirm the VMS supports the intended storage protocol and failover model, calculate sustained write and retention requirements, design redundant paths and power, and agree who will manage the storage over its lifetime. These checks prevent a costly mismatch between design and operation.

A practical checklist for integrators and designers:

  1. VMS compatibility: verify with the VMS vendor which storage types and protocols are supported and how failover recording is handled. Never assume support.
  2. Workload modelling: calculate aggregate bitrate, retention, RAID overhead and analytics metadata, with headroom for growth.
  3. Network design: check that the storage fabric has enough bandwidth, redundant paths and appropriate segregation from camera and corporate traffic.
  4. Single points of failure: review controllers, switches, power and cooling, not just disks.
  5. Support and ownership: decide who monitors, patches and replaces components, and what warranty and response arrangements apply.
  6. Total cost: compare rack space, power, cooling and licensing for the SAN against the equivalent local storage design.

It also helps to question the baseline specification itself. A requirements matrix written around a SAN may hide a simpler design, and one written around local storage may miss a genuine need for shared capacity. Secure Logiq’s system design service exists for this kind of review, covering bandwidth and storage calculations and optimised alternatives for demanding enterprise projects.

If you are weighing a SAN against local storage for a specific project, talk to the Secure Logiq team about your camera count, retention and resilience targets via our contact page, and we can help you test the assumptions before the specification is fixed.

Frequently Asked Questions

How do I get started evaluating whether my CCTV project needs a SAN?

Begin by documenting three numbers: aggregate camera bitrate, required retention in days, and the number of recording servers involved. Then work out the total capacity and sustained write rate, including RAID overhead, hot spares and 20-30% headroom for growth. If a single well-specified server or a few independent servers can meet those figures, local storage is usually enough. If you need pooled capacity or failover between recording servers, shortlist SAN options and validate them with your VMS vendor.

What if my VMS doesn't support the storage protocol or failover model I want to use?

Do not proceed with the SAN design until this is resolved, as an unsupported configuration can leave you without vendor support or with failover that doesn’t work in practice. Ask the VMS vendor for written confirmation of supported protocols (such as iSCSI or Fibre Channel), supported failover recording modes and any licensing needed. If the VMS can’t support shared storage failover, consider alternatives such as mirrored archives, camera-side edge recording or well-specified local storage on each server.

What are the most common mistakes when specifying a SAN for video surveillance?

The most frequent errors are sizing for capacity but not sustained write throughput, forgetting to include analytics metadata in the workload, and leaving single points of failure such as one controller, one switch or a single power feed. Another common mistake is assuming RAID acts as a backup, when it only protects against drive failure. Finally, many projects fail to assign long-term ownership for monitoring, patching and replacing components.

How can I test that my storage will cope before the cameras are installed?

Use a load-simulation tool to generate the expected camera streams against the proposed server and storage, so you can see real throughput limits before physical installation. Secure Logiq’s Logiqal Benchmark utility is designed for this purpose. Run the test with realistic bitrates, include playback and export activity alongside recording, and repeat it with analytics enabled if you plan to use them.

Can I start with local storage and move to a SAN later?

Often yes, but it is much easier if you plan for it at the design stage. Choose a VMS that supports shared storage and failover, leave room in the network and rack design for a storage fabric, and avoid locking archives into formats or layouts that are hard to migrate. Migrating live archives can be disruptive, so if growth is likely to push you towards shared capacity, it’s worth having that conversation before the initial specification is fixed.

Does a SAN remove the need for a separate backup or secondary archive?

No. A SAN with RAID protects against drive and, with the right design, server failure, but it does not protect against accidental deletion, corruption, ransomware or loss of the whole site. If footage is business-critical or evidential, consider a separate archive or offsite copy, along with clear policies for exporting and protecting key incidents.

How do I compare the total cost of a SAN against local storage fairly?

Look beyond the price of disks and include controllers, switches, cabling, licences, rack space, power, cooling, support contracts and the staff time needed to manage the storage. Then price an equivalent local-storage design using high-density recording servers with the same capacity, retention and RAID protection. Comparing both over the expected lifetime of the system, typically 5 years or more, gives a more realistic picture than comparing upfront hardware costs.

Related Articles