Migrating from a Legacy NVR to an Enterprise VMS: Inventory, Parallel Recording and Cutover

Inventory, ONVIF testing, licensing, pilot, parallel recording and a reversible cutover when moving from NVR to VMS; commonly skipped steps and the order of checks.
Technical summary
Technology
NVR, VMS, ONVIF Profile S/T/G, RTSP, NTP
Topic
Carrying out a migration from a legacy NVR to an enterprise VMS
Level
Intermediate–advanced
Prerequisites
  • The VMS platform decision must already be made
  • A camera and NVR inventory
  • A recording retention policy
In this article8 sections

In a move from NVR to an enterprise VMS, choosing the platform is the easy part. What really strains the project is implementation: some cameras don’t send events or PTZ in the new VMS, the license plan didn’t account for the parallel recording period, the old NVR is shut down too early and access to recordings whose retention period hasn’t expired is lost. This picture is common on site. Below we split the migration into inventory, compatibility testing, pilot, parallel recording, cutover and acceptance phases, and write down what goes wrong at each phase and what to check first. If you haven’t chosen a platform yet, read the VMS or NVR comparison first.

Pre-migration inventory and compatibility testing

The input to the migration is a record verified on site for every camera. The NVR’s channel list doesn’t do this job. The NVR may be managing cameras with a proprietary protocol, sometimes on its own internal network via the PoE ports on its back panel; in that case the IP address shown in the list means nothing on your network. The inventory should have at least these fields:

Field Why it’s needed
Brand, model, firmware Checking VMS driver and ONVIF support
IP, MAC, switch port Network migration and addressing plan
Stream profiles Carrying over recording and live viewing settings
Features in use Motion, analytics, I/O, audio, PTZ
Recording mode and retention Storage sizing
Location and purpose Prioritization and wave plan

How to do compatibility testing

At least one sample of each camera model is added to the test VMS in two ways: with the VMS’s manufacturer-specific driver, if there is one, and via ONVIF. The test should give clear answers to these questions:

  • Can the main and substreams be pulled separately, and do the codec and resolution appear correctly?
  • Do motion or analytics events on the camera arrive in the VMS as events? On the ONVIF side, this depends on Profile S/T event support and whether the VMS processes those event topics.
  • Do PTZ, presets, audio and digital I/O work?
  • If transferring the camera’s SD card recording back to the VMS (edge storage) is needed, is it supported? On the ONVIF side, this relates to Profile G.

For the differences between profiles, see the ONVIF Profile T, G and M differences article, and for official definitions the ONVIF profiles page. A conformance declaration shows a model is on the list; it doesn’t show how a specific firmware version behaves with a specific VMS version. This is the common surprise: the stream arrives, the events don’t. Test with the real firmware on site.

The test output is a single decision per model: “full with driver”, “sufficient with ONVIF”, “firmware update needed” or “must be replaced”. Models needing updates are handled through a separate process; see the camera and NVR firmware update plan.

Licensing and server preparation

With enterprise VMS products, the licensing model varies by manufacturer. Common items are camera or channel licenses, client or concurrent session licenses, analytics and integration modules, and annual software maintenance and upgrade fees. When planning, clarify these three points:

  • During the parallel recording period, the same camera records to both the NVR and the VMS; the VMS license is needed from day one.
  • Multi-sensor cameras consume a single license on some VMS products and one per sensor on others.
  • Whether additional licenses are needed for failover recording servers.

Storage is calculated from real stream profiles. The “days remaining” value the NVR shows on screen doesn’t reflect new bitrates; buying disks based on it ends with noticing within the first month that retention has shrunk. For the calculation method, the approach in the How many days does an NVR record? article also applies to a VMS. On servers, NTP, operating system updates, antivirus exclusions (those recommended by the manufacturer documentation for recording directories) and the backup plan must be finished before recording starts. Antivirus without exclusions fills the disk queue every hour it scans the recording directory.

Users, roles and integrations

The second inventory is the people and systems connected to the NVR; it’s often skipped. Who views which camera live, who can export recordings, triggers the NVR receives from the alarm panel or access control system, email and mobile notifications, video wall layouts and remote access methods are gathered in a single list. On NVRs, permissions are often defined per channel and with shared accounts. The migration is the right time to tie roles to Active Directory groups and close shared accounts. For every integration on the list, its counterpart on the VMS side (native integration, ONVIF event, digital I/O or API) should be decided before the pilot. Otherwise, on cutover day part of the alarm flow silently disappears. And nobody notices.

Pilot migration

The pilot is a group of 10–20 cameras including representative points of every camera model and every use type (entrance, perimeter, indoor, PTZ). The goal isn’t only technical functionality; the operator’s workflow is verified too: live viewing layouts, event search, export, user permissions and alarm notifications. The pilot should cover at least one full recording cycle, i.e. night and day conditions and a weekend. Bitrate rises in night images, and a different shift with different habits comes in at the weekend. The pilot isn’t finished until both have been seen.

The parallel recording period

In parallel recording, the camera streams to both the NVR and the VMS. This is what makes the cutover reversible, but it has three constraints:

  • Camera session limit: The number of simultaneous streams a camera can serve is limited. When the NVR, VMS and live viewing clients connect at the same time, the limit can be exceeded; the symptom is usually video coming and going on one client. Check the number of simultaneous users or streams on the manufacturer datasheet, and if necessary route live viewing through the VMS.
  • Network load: When the same stream is pulled twice, uplink and backbone traffic increases. Check the camera network’s capacity against the wave plan.
  • The NVR’s PoE ports: If cameras are connected to the PoE ports on the NVR’s back panel, the VMS can’t reach this internal network directly. First, the cameras are moved to an external PoE switch, and the NVR is set to record them as IP cameras through this switch. The move creates a brief outage per camera; it should be planned separately.

In theory, the parallel period can be extended until VMS recordings reach the target retention period; at most sites this isn’t necessary. The practical way is this: once the VMS’s recording and event behavior meets the acceptance criteria, recording on the NVR is stopped, and old recordings are preserved separately.

Access to old recordings and the retention period

Recordings on the NVR must remain accessible for incident review and data subject requests until the retention period expires. There are three options:

Method Advantage Limitation
Running the NVR in isolation Simple, no tools needed Risk of device failure, separate permission management
Exporting critical recordings An independent archive Exporting everything takes time and space
The VMS adding the NVR as a device Access from a single interface Not every VMS supports it; testing needed

An NVR left in isolation is the device most likely to be forgotten: it gets no updates, its disk silently fails, the person who knows its password leaves. Whichever method is chosen, users accessing the old system are restricted, access is logged and at the end of the retention period the disks are disposed of according to the destruction procedure. For setting the retention period, see the camera recordings and KVKK retention period guide.

Cutover and rollback plan

Cutover is done in waves by zone or building. The order is the same in each wave:

  1. Verify that the cameras in the wave are recording in the VMS and that events arrive.
  2. Enable operator viewing layouts and user permissions in the VMS.
  3. Stop recording these cameras on the NVR; existing recordings on the NVR aren’t deleted.
  4. Redirect alarm, access control and event notification integrations to the VMS.
  5. Collect operator feedback over one shift.

The rollback list is written before the cutover, not after:

  • Recording on the NVR can be restarted; channels weren’t deleted, only disabled.
  • The user account and password the cameras use for the NVR weren’t changed, or the new value is documented.
  • If camera IP addresses were changed, the old address plan and NVR channel mapping are recorded.
  • A criterion for the rollback decision is defined (for example, recording failing in a critical area) and the person who will make the decision is known.

The last item is the one most often skipped. If it’s unclear who will say “roll back” when a problem arises at midnight, the rollback plan stays on paper.

Post-migration acceptance checks

  • No gaps on the recording timeline for any camera; the recording profile matches the plan.
  • Recording server disk usage is progressing in line with the calculated value.
  • All servers and cameras take their time from the same source.
  • User roles are defined on the principle of least privilege; no shared accounts.
  • Export and verification (if signed export is supported) were tested with the operator.
  • If there’s a standby recording server, failover was actually tried.
  • Inventory and network documentation updated to the new state.

Don’t tick the failover item on paper. An untested standby server isn’t a standby.

For architecture and licensing decisions, see our enterprise VMS platforms page, and for a redundant recording architecture, the NVR/VMS high availability page.

Frequently asked questions

Can recordings on the old NVR be moved to the new VMS?

Since recording formats are usually manufacturer-specific, direct transfer is often not possible. Practical ways are running the NVR in isolation for the retention period, exporting critical recordings, or, if supported, adding the NVR to the VMS as a device.

Does parallel recording strain the cameras?

The number of simultaneous streams a camera can serve is limited. When the NVR, VMS and live viewing clients connect at the same time, this limit can be exceeded; checking the manufacturer datasheet and routing live viewing through the VMS reduces the risk.

Does every ONVIF-compliant camera work smoothly with the VMS?

No. An ONVIF conformance declaration shows the profile a model supports; it doesn’t guarantee the event, PTZ or edge recording behavior of a specific firmware with a specific VMS version. Testing must be done with the firmware on site.

How long should a pilot migration last?

The pilot should cover at least one full recording cycle, i.e. night and day conditions and a weekend. Operator workflows are verified as well as technical functionality.

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