Surveillance Disk Selection: Workload Rating, CMR/SMR and Compatibility

Disk selection for NVRs and recording servers: workload rating calculation, the CMR/SMR difference, vibration, SCT ERC, SMART monitoring and the manufacturer’s disk compatibility list.
Technical summary
Technology
Surveillance HDD, CMR, SMR, SMART, SCT ERC, smartctl
Topic
Choosing the disk class for video recording
Level
Intermediate
Prerequisites
  • The recorder’s number of disk slots and RAID layout
  • Total recording bitrate
In this article8 sections

The disk is the part of a recorder that is replaced most often and fails most often. Two disks of the same capacity in the same price band can be completely different in workload rating, recording technology and firmware behavior. Here we explain what to look for on the datasheet when choosing disks for an NVR or VMS recording server, the difference between CMR and SMR, and the pitfalls in purchasing and warranty processes.

The load video recording puts on disks

In enterprise camera systems, a recording disk runs all 8,760 hours of the year and writes for nearly all of that time. Unlike the disk in an office computer, it has no idle period. The impact on disk selection comes down to three parameters: annual operating hours, the amount of data transferred per year (workload rating) and the number of simultaneous streams.

To estimate the workload roughly, use the average transfer rate per disk. A sustained transfer of 1 MB/s amounts to 1 MB/s × 31,536,000 s ≈ 31.5 TB per year.

Worked example: 64 cameras at an average of 4 Mbps each; 256 Mbps in total, or 32 MB/s of writes. In an eight-disk RAID 6, data is written to six disks and parity to two; total physical writes including parity are 32 × 8/6 ≈ 42.7 MB/s, or about 5.3 MB/s per disk. That means about 168 TB of writes per disk per year. Playback and export reads, rebuilds and consistency checks come on top. The result is just below the limit on many surveillance disks’ datasheets. There’s no margin. For a system at this scale, consider a higher workload class.

Disk class comparison

The table below outlines value ranges commonly found in manufacturer datasheets. Exact values differ for each model and capacity; check the current datasheet of the relevant model before buying.

Class Design assumption Typical workload rating For video recording
Desktop Limited operating hours, single disk Low (e.g. 55 TB/year) Not suitable
NAS 24/7, multi-disk enclosure 180–300 TB/year Usable if compatible
Surveillance 24/7, write-heavy, multiple streams 180 TB/year, 550 TB/year in the top series The main option for NVRs
Enterprise 24/7, high slot count, heavy I/O 550 TB/year Large recording servers and storage arrays

What really sets the surveillance class apart from the NAS class is firmware behavior. These disks are tuned for many simultaneous write streams. Some models, together with recorders using the ATA Streaming command set, prioritize responding in time over retrying for a long time on a read error. In video, a few corrupted frames do far less harm than a disk going silent for seconds and the recording queues of all streams filling up.

Workload rating and continuous operation

The workload rate is the annualized total of data written to and read from a disk in a year; manufacturers tie it to warranty terms. Reads count toward this value, not just writes. When choosing:

  • Calculate the annual load per disk using the method above, and leave margin for rebuilds, playback and export in recording storage capacity planning.
  • If the calculated value is close to the limit of the class you’ve chosen, move up a class or increase the number of disks in the array to lower the load per disk.
  • Check the annual operating hours on the datasheet. On some desktop disks this value is limited to a few thousand hours a year; 24/7 use falls outside the assumption.
  • Monitor operating temperature. Recorders often sit in poorly ventilated cabinets. A common picture is disks pushed to their upper temperature limit in a wall cabinet with its door closed; that’s where lifespan gets cut.

The difference between CMR and SMR

On a CMR (Conventional Magnetic Recording) disk, tracks don’t overlap; any sector can be rewritten directly. On an SMR (Shingled Magnetic Recording) disk, tracks are written partly overlapping, like roof shingles. More capacity fits on the same platter, but a change in one area requires neighboring tracks to be rewritten too.

On drive-managed SMR (DM-SMR) models, the disk first collects incoming writes in a fast CMR cache zone and moves them to the main zones when idle. In video recording, the disk is never idle. When the cache fills, write performance drops markedly and latency becomes unpredictable. A RAID rebuild requires uninterrupted writing across full capacity; on an SMR disk it takes very long and can cause the controller to drop the disk from the array on a timeout.

Vibration and multi-disk enclosures

In a multi-disk enclosure, each disk transmits vibration to its neighbors as it spins. Disks designed for single-disk use lack the sensors and servo correction to compensate for this rotational vibration. Head positioning errors turn into retries, and retries into lower write performance.

  • The datasheet states the maximum number of supported bays. A disk approved for up to 8 bays isn’t the right class for a 16- or 24-bay recording server.
  • In servers and JBOD enclosures with many bays, enterprise-class disks are designed for vibration tolerance.
  • Fitting all disk sleds and screws properly and mounting the enclosure on a stable rack are simple but often skipped measures.

In systems using RAID, the disk’s error recovery time is also decisive. Desktop disks may retry for a long time on an unreadable sector; meanwhile the controller may consider the disk unresponsive and drop it from the array. On NAS, surveillance and enterprise disks this time is limited (SCT Error Recovery Control). On a Linux-based recording server, you check the status like this:

# Disk model, serial, health summary and SMART attributes
smartctl -a /dev/sdb

# Read the SCT Error Recovery Control value (unit: 100 ms)
smartctl -l scterc /dev/sdb

# If supported, set read/write recovery time to 7 seconds
smartctl -l scterc,70,70 /dev/sdb

On some disks this setting reverts to the default after a power cycle; if it needs to persist, add it to a startup script. In SMART output, watch for increases in the reallocated sector (5), pending sector (197), uncorrectable sector (198) and interface CRC error (199) attributes. If CRC errors are rising, look first at the cable and backplane, not the disk. The disk is often innocent.

Manufacturer compatibility lists

NVR and recorder manufacturers publish the disk models they have tested and approved in compatibility lists. This list isn’t just “it works” information; it shows that the device’s disk health monitoring, sleep management and error behavior have been verified with that disk.

  • A disk not on the list often works; however, the device manufacturer’s support may be limited in case of a failure.
  • The maximum supported disk capacity depends on the device model and firmware version. Read the firmware notes before moving to a larger disk.
  • On enterprise recording servers, the server manufacturer’s supported disk list applies; third-party disks can trigger warnings on the RAID controller or affect warranty coverage.

With which protection scheme the disks are put into an array is a separate decision; for that, see the RAID level selection for video recording guide. Total capacity needs are determined by the NVR storage calculation.

Purchasing checklist

  • Was the annual workload per disk calculated and compared with the chosen model’s workload rating?
  • Was the recording technology verified as CMR by model number?
  • Is the number of supported bays on the datasheet equal to or greater than the enclosure’s number of bays?
  • Is the model on the recorder’s or server’s compatibility list?
  • Are all disks from the same batch? In large arrays, buying disks from different shipments reduces the risk of simultaneous wear.
  • Have the warranty period and the warranty’s validity under 24/7 use been confirmed?
  • How will data security be ensured when returning a failed disk? Footage on the disk is personal data; assess in advance the options for destroying an unreadable disk instead of returning it, and keep the process consistent with the camera recording destruction procedure.
  • Is a stock of spare disks kept? The time between a failure and procurement adds to the time the array stays unprotected.
  • Are SMART monitoring and disk failure alarms enabled on the recorder or monitoring system?

If you’re seeing frequent disk failures or capacity problems in existing recorders, we address disk class, array layout and monitoring together under our NVR disk failure and capacity upgrade service. The official resource for using SMART tools is the smartmontools project page.

Frequently asked questions

Can a NAS disk be used instead of a surveillance disk?

In most cases it can; NAS disks are designed for continuous operation and multi-disk enclosures. Three things decide: the workload rating on the datasheet must cover the calculated load, the disk must be CMR, and it must be on the recorder’s compatibility list.

Why do desktop disks fail early in recorders?

Desktop disks are designed on the assumption of limited annual operating hours and a low workload rating; they lack sensor and firmware support for vibration in a multi-disk enclosure. Year-round continuous writing falls outside these assumptions and can also affect warranty terms.

How do you tell whether a disk is CMR or SMR?

The most reliable source is the manufacturer’s current datasheet or recording technology list. Check by model number; different capacities in the same product family may use different recording technologies.

What happens to recordings when a failed disk is sent for warranty?

Footage on the disk contains personal data. If the disk is readable, securely erase it before returning it. If it isn’t readable, use a warranty option that allows destruction without return, or an internal destruction procedure.

Support

Contact us for support or a discovery session on this topic.

Contact form
[email protected]

Eretek Video Surveillance

IP cameras, recording and central management infrastructure. We plan blind spots, retention periods and the camera network around the risks of each site.

Knowledge Center

Other articles