Lead Forensics

How does SAN storage differ from direct-attached storage in a VMS deployment?

A storage area network (SAN) provides a shared pool of recording storage that multiple VMS recording servers access over a dedicated network, while direct-attached storage (DAS) connects drives to a single server. DAS is simpler and cheaper to deploy, whereas SAN storage adds central management, flexible capacity allocation and failover options for larger surveillance storage estates.

The right choice depends on scale, resilience requirements and how many recording servers share the same estate. The sections below explain how video travels in each design, where they differ, and when it makes sense to move from one to the other.

How does video data travel to SAN versus DAS?

With DAS, video streams from cameras reach the recording server over the network, and the server writes them straight to its own local disks through an internal or external storage controller. With SAN, the recording server first receives the streams, then writes them across a separate storage network to shared storage arrays.

In a DAS design, the path is short: camera, switch, recording server, local RAID array. The server owns its storage, and the VMS sees it as a local volume. Few components sit between the stream and the disk, which keeps latency low and troubleshooting straightforward.

In a SAN design, the recording server still ingests the camera streams, but the disks live elsewhere. Writes travel over a dedicated storage network, typically using block-level protocols such as iSCSI or Fibre Channel, to arrays that several servers can use. The VMS still treats the storage as a local disk, because SAN presents block-level volumes to the operating system.

That extra hop matters for sustained write workloads. Surveillance recording is continuous, sequential and write-heavy, so the storage network needs enough bandwidth and consistent latency to avoid dropped frames. Video that spans many cameras can generate considerable aggregate throughput, so the SAN fabric must be designed around peak, not average, load.

What are the main differences between SAN and DAS?

The main difference is sharing. DAS dedicates storage to one recording server, while SAN pools storage that many servers can access. That changes how you scale, manage and protect video data, and it introduces extra components, cost and design complexity in return for flexibility.

Comparing the two on consistent criteria makes the trade-offs clearer:

  • Architecture: DAS ties disks to a single server. SAN separates compute from storage using a dedicated storage network.
  • Scalability: DAS grows by adding drives or enclosures to one server, which eventually hits chassis and controller limits. SAN scales capacity and recording servers more independently.
  • Capacity utilisation: DAS often leaves stranded capacity on lightly loaded servers. SAN lets you allocate volumes from a common pool.
  • Management: DAS is managed per server. SAN centralises provisioning, monitoring and firmware management.
  • Complexity and cost: DAS has fewer parts and a lower entry cost. SAN adds switches, controllers and design effort.
  • Performance behaviour: DAS offers a short, predictable path. SAN performance depends on the whole fabric and on how contention is managed.

Neither is inherently better. A well-designed DAS system can be more reliable in practice than a poorly designed SAN, because every added component is another place for something to fail. The question is whether the benefits of shared storage justify that added complexity for your deployment.

Does SAN improve resilience and failover for VMS?

SAN can improve resilience and failover because recording data sits independently of any single recording server. If a server fails, another can be pointed at the same storage volumes, depending on how the VMS handles failover. However, a SAN also introduces new single points of failure that must be designed out.

The key benefit is decoupling. With DAS, a failed server can strand its recordings until it is repaired, unless the VMS replicates data elsewhere. With a SAN, the video remains on shared arrays, so a standby or replacement recording server can take over its workload. Exactly how this works depends on the VMS platform and its failover feature set, so always check the vendor’s supported architectures.

Resilience is not automatic, though. Consider the following:

  • The SAN itself: A single storage controller, switch or network path can take down every server that depends on it. Redundant controllers, dual fabrics and multipath connectivity are standard design practice for this reason.
  • RAID is not backup: RAID protects against drive failure, not deletion, corruption or site loss. Redundancy also does not guarantee uninterrupted recording, so hot spares and rebuild behaviour need to be planned.
  • Failover testing: A failover design that has never been tested is an assumption, not a safeguard.

For critical environments, resilience is as much about design discipline as technology choice. System design that maps every failure scenario to a defined outcome is what turns shared storage into genuine availability.

Is DAS still enough for smaller VMS deployments?

Yes. For small and medium VMS deployments, DAS is often the more sensible choice. A single purpose-built recording server with hardware RAID, hot spares and redundant power can deliver high reliability without the cost and operational overhead of a storage network.

DAS suits deployments where the camera count and retention period fit comfortably within one server’s capacity, where there is no requirement for automatic failover between servers, and where the team prefers a simple, easily supported architecture. Fewer components also mean fewer things to monitor and maintain, which matters at sites without dedicated IT storage expertise.

DAS does have limits. Capacity is constrained by the chassis and controller, expansion tends to be lumpy, and spare capacity cannot easily be shared with other servers. If a recording server fails, restoring service usually means repairing or replacing that server, so a clear spares and support plan matters. Hardware designed specifically for surveillance, such as the HD surveillance servers Secure Logiq builds, addresses many of these concerns at the server level through dedicated OS drives, hardware RAID and redundant power.

When should a VMS design move to SAN storage?

A VMS design should move to SAN storage when the number of recording servers, the volume of retained video or the availability requirements make isolated server storage inefficient or risky. Typical triggers include multi-server estates, large retention requirements, shared failover expectations and centralised management needs.

Consider SAN when several of these conditions apply:

  • Many recording servers: Pooled storage reduces stranded capacity and simplifies growth across a large camera estate.
  • Long retention at high bitrates: Capacity requirements that outgrow a single server’s drive bays push designs towards shared, expandable arrays.
  • Failover expectations: If the specification requires another server to resume recording against existing storage, shared storage is often part of the answer.
  • Centralised operations: Estates managed by a central IT or security operations team benefit from unified storage monitoring and provisioning.
  • Analytics growth: AI analytics add metadata and processing load, which can change storage behaviour and growth patterns.

Equally, do not move to SAN simply because the project is large. Validate the throughput, retention and failover requirements first, using accurate bandwidth and storage calculations, then decide whether shared storage delivers a measurable benefit over well-specified DAS. For a real-world example of an enterprise project using this approach, see the Qatar University SAN storage case study.

If you are weighing SAN against DAS for a specific project, the Secure Logiq system design team can review your camera counts, retention targets and resilience requirements and recommend an architecture that fits.

Frequently Asked Questions

How do I calculate how much storage my VMS recording servers need before choosing SAN or DAS?

Start with the number of cameras, the bitrate per stream (based on resolution, codec, frame rate and scene complexity), the recording mode (continuous or motion-based) and the required retention period. Multiply average bitrate by recording time to get capacity, then add headroom for RAID overhead, metadata and growth. Most VMS vendors provide storage calculators, and you should size throughput around peak load, not just average, before deciding whether DAS can cope or shared storage is justified.

What network design should I use for iSCSI or Fibre Channel storage in a surveillance SAN?

Keep storage traffic on a dedicated network, separate from camera and client traffic, so recording writes are never competing with other workloads. Use redundant switches, multipath connectivity and enough bandwidth for peak aggregate write throughput. For iSCSI, dedicated switches or isolated VLANs with appropriate speeds and jumbo frames (where supported end to end) are common practice, while Fibre Channel typically uses dual fabrics for redundancy.

Can I start with DAS and migrate to SAN later as my system grows?

Yes, but plan for it early. Choose recording servers that support the connectivity you may need later, such as additional network ports or host bus adapters, and use a VMS that supports flexible storage paths. When migrating, you will typically add the new storage as an additional recording location, let new footage write there, and allow older DAS-based footage to age out or be moved according to your retention policy and vendor guidance.

What are the most common mistakes when deploying SAN storage for video surveillance?

Common mistakes include sizing for average rather than peak throughput, creating single points of failure in switches or controllers, and sharing the storage network with general IT traffic. Others are skipping hot spares and rebuild planning, and assuming failover will work without testing it. Using enterprise SAN sizing assumptions designed for random-access workloads, rather than continuous write-heavy video, can also lead to dropped frames.

How should I test failover on a SAN-based VMS deployment?

Test each failure scenario in a controlled maintenance window: power down a recording server, pull a storage network path, fail a controller and simulate a drive failure and rebuild. For each test, confirm that recording resumes, note how long the gap in video is, and check that alerts reach the right people. Document the results and repeat the tests after major firmware, VMS or configuration changes.

If I stay with DAS, how can I make a single recording server as resilient as possible?

Use hardware RAID with hot spares, redundant power supplies and a dedicated OS drive separate from the recording volumes. Add proactive monitoring for drive health and RAID status, keep spare parts or a support agreement with fast replacement, and consider VMS-level features such as failover recording servers or edge recording on cameras. Remember that RAID is not a backup, so protect critical footage separately if it must survive deletion or site loss.

Does adding AI analytics change whether I should choose SAN or DAS?

It can. Analytics generate additional metadata, may require extra processing and can change how much video you retain or export, which affects both storage capacity and I/O patterns. Re-run your capacity and throughput calculations with analytics enabled and include growth headroom. If analytics expansion is likely to push you beyond a single server’s drive bays or require pooled storage across several servers, that strengthens the case for SAN.

Related Articles