The software from the people who make the cameras. And it shows in everything.

The gap it fills: video properly managed across several industrial units or several buildings without setting up a department to maintain it.

Six things that decide the sizing, and one of them decides the migration too.

• There is no central server everything depends on: each server keeps the configuration and can serve the client, so if the one you use goes down you come in through another. That takes away the single point of failure other platforms solve with a cluster and its licence.
• You grow by adding machines and merging systems: each site is built complete and then added in, sharing users, permissions and views. If one machine goes down, another picks up its cameras — with the drives and the capacity worked out beforehand for its own load and for somebody else's, which is the part that does not happen by itself.
• Recording goes to local drives, an array or a NAS, with separate backup storage. And the setting that decides retention most is not the drive: it is recording the small stream all the time and the big one only when there is movement. It turns days into weeks, and the price is detail in the hours when nothing happened.
• The archive stays tied to the server that wrote it. It gets discovered late and it has a very concrete consequence: you can move a camera to another machine, its history stays where it was. Merging three servers into one means keeping the old ones switched on or exporting, and that time gets budgeted.
• Its cloud is the door, not the warehouse: it is there so you can come in from outside without opening ports and so a user can sign in with their account. The video stays on your drive. What does leave the house are the accounts and the connection data, which are personal data even though they are not video — and you can work with no cloud, with local users and a VPN, and then remote access is IT's job.
• Access control and intruder detection are integrated, they do not come inside: the panel raises the alert, WAVE marks the video, opens it on the operator's screen and records at top quality. The limit is clear and has to be said — the card, the cardholder and the audit report go on living in the other platform. And there is a development kit and a programming interface, inherited from the base; which specific integration is supported on this brand is asked of the manufacturer in writing.

Both things are true and neither is in the brochure in full. It is worth having them together before deciding the camera estate, because that decision is the one that rules.

The question that decides is not the size. It is whose cameras these are going to be in five years' time.

Design, sizing, deployment, commissioning, migration and maintenance. Put like that it says nothing, so here it is in concrete decisions.

• The sizing written down: how many servers, how much drive space and how many days of real retention with the configuration handed over —dual stream included—, not with the brochure's. The calculation is handed over, not the total, so it can be argued with.
• The network before the software: video kept separate from the company's network, cabinets, power, backup power and where the traffic goes in the hour when everybody is watching.
• The test rig with the cameras that are really there, of every brand there is, checking which functions arrive and which do not. It is the test that turns «it is compatible» into a list, and we do it even when it spoils our own quotation.
• The deployment: one system per site and the merge afterwards, permissions by profile and not by person, and the failure recovery tested by switching a machine off in front of the client. A system sold as redundant that nobody has switched a server off on is not tested, it is configured.
• The rules, which is where this stops being a recorder: what event opens what camera, who gets the alert at four in the morning and what gets recorded at top quality. With the access control or intruder detection integration tested by triggering the event on the real system.
• The migration and the maintenance: living alongside the old system for a few weeks, exporting what has to be kept before switching it off, and after that drives, versions, fields of view, retention measured properly and an alert when a camera stops recording. A video system degrades in silence.

It splits well between two teams. The split gets decided before handover: afterwards it decides itself, and it usually decides badly.

• IT keeps the machines, the drives, the backups, the accounts, the updates and the camera network. It is a service on Windows or Linux, so anybody who already maintains servers learns nothing strange — and that is a real advantage for a department of two people.
• And it keeps a task that here is theirs and on other platforms does not exist: managing the camera estate. Firmware, passwords and certificates of a single brand are a perfectly normal IT task, and if nobody takes it on the estate grows old on the factory password.
• Security keeps the views, the bookmarks, the searches, the alarms and the exports. Who can take images out, with what authorisation and what gets logged is a data protection decision and a matter of judgement, not an IT one: in IT's hands it gets settled by whatever is easiest to configure.
• What is not left to daily use: the retention and the recording quality. They are two boxes and they move terabytes, so that access is limited in writing and by name, not by custom.
• It asks little of the installer and quite a lot on rules, on the API and on knowing which camera brings what. Building this is easy, and that is why there are WAVE installations that are an expensive recorder: what turns it into a system is in the part you do not see.
• Training on two levels: half a day for whoever watches, searches and exports —and to know what to do when an image does not appear—, and a technical session with IT on architecture, backups and what to do when a server does not come back. Documentation: a diagram of the servers and which camera records on which, a table of what analytics each model brings, the retention calculation, the rules in plain language and the administrator passwords in the client's name.

Here the analytics are not bought separately: they arrive inside the cameras. That has one big advantage and one very concrete limit.

If you are between this one and an open platform, the underlying question is not the function list: it is how many camera brands you are going to have in five years' time. Both answers can be defended.