A SAN storage specification for enterprise CCTV should state usable capacity against retention and bitrate, sustained sequential write throughput, RAID and failure tolerance, network connectivity and protocols, VMS certification, and warranty and support terms. Each requirement should be measurable, so suppliers can be compared like for like. This guide covers each element in turn and how it applies to surveillance servers writing to shared SAN storage.
How much SAN capacity does an enterprise CCTV system need?
SAN capacity is calculated from the number of cameras, the recorded bitrate of each stream, the required retention period and a margin for growth and overheads. The specification should state usable capacity after RAID and formatting overheads, not raw drive capacity, because the two figures can differ significantly.
The basic calculation is: bitrate (Mbps) multiplied by seconds recorded, divided by eight, per camera, summed across the estate. Bitrate depends on resolution, codec, frame rate, scene complexity and whether recording is continuous or motion-triggered. Always use the bitrate your cameras will actually produce, not a datasheet maximum.
A specification should also account for:
- Retention policy: The required number of days, and whether any cameras or events need longer retention.
- Growth: Planned camera additions and any resolution upgrades during the system’s life.
- Analytics metadata: AI analytics can generate substantial metadata and secondary streams in addition to video.
- Headroom: Free space reserved so the array is not running at full capacity, which can degrade performance.
- Expansion path: Whether capacity can be added through additional enclosures without rebuilding the array.
Treat any worked example as an assumption-based estimate, and ask your supplier to show the bandwidth and storage calculations behind their proposal. For a deeper treatment of server-side sizing, the guide to selecting a server covers the surrounding considerations.
What throughput and IOPS should a CCTV SAN deliver?
A CCTV SAN should be specified for sustained sequential write throughput that comfortably exceeds the combined bitrate of all recording streams, plus playback, export and any analytics reads. IOPS matter less than for databases, because video is mostly large, sequential writes.
Start with the aggregate recording bandwidth across all cameras, then add the concurrent read load from live viewing, investigations, exports and analytics. The specification should require the SAN to sustain that combined load continuously, not just in short bursts from cache.
Points worth stating explicitly:
- Sustained, not peak, performance: Cache can mask weak drive performance for a short time. Ask for figures under continuous 24/7 load.
- Performance during degradation: Require stated performance while the array is rebuilding after a drive failure, since this is when frames are most at risk.
- Latency under concurrency: Multiple recording servers writing at once can behave differently from a single host.
- Mixed workloads: Live recording, playback and analytics compete for the same spindles and controllers.
Storage performance depends on the whole architecture, including controllers, network, host bus adapters and recording server configuration, not just the drives. Where possible, validate the design before deployment. Secure Logiq’s Logiqal Benchmark utility, for example, lets designers simulate camera resolutions, compression formats, frame rates and analytics workloads to test throughput limits ahead of physical installation.
Which RAID and resilience features should the specification list?
The specification should name the RAID level, hot spare provision, controller redundancy, power redundancy and drive class. For continuous video recording, parity-based levels such as RAID 5, 6 or 60 with hot spares are common, with RAID 6 or 60 favoured for large arrays because they tolerate multiple drive failures.
Resilience should be specified at every layer, not only at the drives:
- RAID level and failure tolerance: State how many simultaneous drive failures the array must survive.
- Hot spares: Require dedicated or global hot spares so rebuilds start automatically.
- Controller redundancy: Dual controllers or equivalent, so a controller fault does not stop recording.
- Power supplies: Redundant, hot-swappable units on separate power feeds where possible.
- Drive class: Enterprise drives rated for 24/7 surveillance workloads rather than desktop-class drives.
- Hot-swap capability: Drives, fans and power units replaceable without shutting down.
- Multipath connectivity: Redundant paths between recording servers and the SAN.
Bear in mind that RAID is not a backup, and redundancy does not guarantee uninterrupted recording. If your risk assessment requires protection against site loss, corruption or accidental deletion, the specification should also address replication, archive or failover recording separately.
What network connectivity and protocols should a SAN specify?
The specification should define the host interface type, link speeds, number of redundant paths and supported protocols, and require that these match the recording servers and switching infrastructure. Common options include Fibre Channel and iSCSI, and the right choice depends on the existing environment and IT policy.
Items to state clearly:
- Protocol: Fibre Channel, iSCSI or another block protocol, based on what the client’s infrastructure and skills support.
- Link speed and port count: Enough bandwidth per host and per array, with headroom above aggregate write and read load.
- Redundant fabrics or networks: Separate paths and switches so a single switch failure does not interrupt recording.
- Network separation: Whether storage traffic runs on a dedicated network, isolated from camera and corporate traffic.
- Management access: How the array is monitored and managed, including alerting and secure access.
SAN is not inherently the best choice for every CCTV deployment. Direct-attached storage or server-based storage can be simpler and cheaper for smaller or segmented systems, while a SAN suits large estates needing shared, scalable capacity. Do not assume a particular protocol or feature is available from any supplier without confirming it in their documentation.
Does the SAN need to be certified for your VMS?
Yes, it is strongly advisable. VMS vendors publish requirements and, in some cases, validate hardware platforms, and certified or validated storage reduces the risk of compatibility, performance and support disputes. At minimum, the specification should require written confirmation that the proposed SAN design is supported for the chosen VMS.
Check what the VMS vendor actually supports, including recording server architecture, storage types, file system and any recommended configurations. Some VMS platforms have specific guidance for networked storage, and unsupported designs can complicate technical support.
The specification should ask suppliers to:
- Name the VMS and version the design has been validated against.
- State whether the hardware is certified, validated or simply compatible, since these terms are not interchangeable.
- Describe how the design handles recording server failover, if the VMS supports it.
- Confirm any integration with analytics platforms that read from the same storage.
Secure Logiq builds servers that are optimised, tested and certified for leading VMS platforms, and works with vendors including Milestone, Genetec and others. For SAN-specific designs, confirm the exact certification status for your VMS and version directly. A real example of shared storage in practice is the Qatar University SAN storage case study.
What warranty and support terms belong in the specification?
The specification should state warranty length, response time, whether cover is on site, which components are included and how long replacement drives take to arrive. For 24/7 recording systems, next business day on-site cover over the system’s full expected lifespan is a sensible baseline to request.
Terms worth writing into the specification:
- Warranty duration: Match it to the planned operational life of the system, commonly several years.
- Response model: Advance replacement versus on-site engineer, and the guaranteed response time.
- Component coverage: Confirm drives, controllers, power supplies and chassis are all included, not only some parts.
- Drive retention: Whether failed drives can be retained by the client, which matters for sensitive footage and data protection policy.
- Technical support: Where support engineers are based, hours of availability and escalation routes.
- Proactive monitoring: Alerting for drive health, temperature and array status.
Secure Logiq offers a minimum three-year advance replacement warranty on its products and a five-year next business day on-site warranty on enterprise products, with support provided from its UK headquarters. Treat any supplier’s warranty terms as contractual details to verify in writing rather than assume.
If you are drafting a SAN specification for a large surveillance project, Secure Logiq’s system design service can help validate capacity, throughput and resilience calculations before you go to tender.
Frequently Asked Questions
How do I get started drafting a SAN specification for a CCTV tender?
Begin by collecting the inputs every other requirement depends on: camera count, actual recorded bitrate per stream, retention days, planned growth and your chosen VMS and version. Then turn each section (usable capacity, sustained write throughput, RAID and failure tolerance, connectivity, VMS certification and warranty) into a measurable, pass/fail line item. Ask bidders to submit their bandwidth and storage calculations alongside their proposals so you can compare them like for like.
What headroom should I build in so the SAN doesn't run at full capacity?
There is no single universal figure, but you should specify free space reserved above your calculated retention requirement, plus allowances for camera additions, resolution upgrades and analytics metadata. Arrays running close to full can see degraded performance, so ask suppliers to state their recommended maximum utilisation and show how it was factored into the usable capacity figure. Also confirm that capacity can be expanded with extra enclosures without rebuilding the array.
How can I verify a supplier's throughput claims before the system is installed?
Ask for sustained sequential write figures under continuous 24/7 load rather than cache-assisted peaks, and request performance data during an array rebuild. Where possible, simulate your real workload before deployment using a benchmarking tool, such as Secure Logiq’s Logiqal Benchmark, which can model resolutions, codecs, frame rates and analytics loads. A pilot or proof-of-concept test with multiple recording servers writing at once is also worthwhile, as concurrency can behave differently from a single host.
What happens to recording if a drive fails or the array is rebuilding?
With a suitable RAID level (such as RAID 6 or 60), hot spares and redundant controllers, the array should keep recording while a failed drive is rebuilt, but performance may drop during that period. This is why the specification should require stated throughput during degradation and rebuild. Remember that redundancy does not guarantee uninterrupted recording, so test failure scenarios during commissioning and set up proactive alerting for drive health.
Is RAID enough, or do I also need backup, replication or failover recording?
RAID protects against drive failures, not against site loss, data corruption, accidental deletion or controller-level problems. If your risk assessment calls for protection against these, specify replication, archive storage or failover recording separately from the primary SAN. Decide which footage, if any, justifies the extra cost, such as high-risk cameras or events needing longer retention.
Should I choose Fibre Channel or iSCSI for my CCTV SAN?
The right answer depends on your existing infrastructure, IT policy and in-house skills rather than one being universally better. Fibre Channel suits environments that already have a fabric and expertise, while iSCSI can use standard Ethernet switching but benefits from a dedicated, isolated storage network. Whichever you choose, specify redundant paths and switches, enough link speed with headroom above combined read and write load, and confirm protocol support in the supplier’s documentation.
What should I do if the VMS vendor doesn't list my proposed SAN as certified?
First clarify whether the hardware is certified, validated or merely compatible, as these terms are not interchangeable. Ask the supplier for written confirmation that the design is supported for your VMS and version, and check the VMS vendor’s published guidance on networked storage, file systems and recording server architecture. If support is unclear, consider an alternative design or have the supplier validate it before tender, since unsupported configurations can complicate technical support later.
Related Articles
- How can surveillance SAN capacity be expanded without interrupting recording?
- How can CCTV archives be protected against accidental deletion or corruption?
- How can CCTV footage be replicated between separate sites?
- What are the signs that your surveillance server is undersized?
- Should you choose a surveillance server or an NVR for enterprise use?