European, on your servers, and without tying you in.

There are projects where the choice does not start with functions. It starts with three conditions.

Classic server architecture, with a separation you notice while working: administering and operating are two different tools.

• There is one server that governs the configuration and, underneath it, the servers that record. That has a consequence for the project: that server needs protecting and backing up. Recording does not stop if you lose it, but administration does, and recovering a system without a copy of its configuration is a long afternoon.
• Administration and operation are kept apart: the administrator adds cameras, users, profiles and storage from their tool; the operator works in theirs —Spotter— with their views, their maps and their exports. That lets you give plenty of use without giving away any configuration.
• Scaling: you add recording servers according to cameras and retention, and the whole thing is still one system with a single login. For several buildings with their own network between them it is comfortable; for scattered sites with a poor connection, the cloud platforms on this list fit better, and that has to be said.
• Recording and redundancy: drives or an array, a copy to a second storage of whatever has to be kept twice, and failover servers so recording carries on if a machine goes down. It is sizing, not a tick-box: it is calculated beforehand and tested by switching a machine off at handover.
• Access from outside: a gateway for web and mobile, with the network separated and a VPN. Here there is no manufacturer service acting as the door, and that means two things at once — nothing leaves your installation unless you decide it does, and there is IT work somebody has to do.
• SDK, integration and licence: there is a development kit and there are interfaces to integrate access control, intruder detection, till points or your own systems. The licence is per channel and bought once; the maintenance agreement —new versions and support— is renewed separately, and not renewing it switches nothing off: it leaves you where you were.

This platform does not have the cloud conversation that five of the seven we write up have. It has another one, and it is shorter.

It has a very clear place, and outside that place there are better options. Saying so is part of the job.

A platform on your servers puts the work back where it was: the project, the calculation and the maintenance.

• The sizing: how many recording servers, what drives, what array if one is needed, how many real days of retention with the configuration as handed over, and what happens to that number when next year's cameras come in.
• The network before the software: video separated from the company network, bandwidth between buildings, cabinets, power and backup. In an installation with several buildings, the network decides half of the problems to come.
• The rollout: servers, a copy of the configuration from day one, adding cameras of every brand there is, profiles by role, recording schedules and the failover servers tested by switching a machine off in front of you.
• The operation, which here is designed: the views each role needs, the maps, which event opens which camera and who gets the alert at four in the morning. A well-sized system with badly thought-out views is used badly every single day. Plus the integrations with access control and intruder detection, tested in both directions.
• The migration and the maintenance: running alongside the old system, exporting whatever has to be kept from the old archive, and after that controlled updates, drives, configuration backups, fields of view and measurement of the real retention. Measured, not estimated: they are two different numbers and the second is always more optimistic.

The split is the classic one, and the separation between administering and operating that the platform brings helps a good deal in getting it right.

• IT: servers, drives, backups —including the one of the system configuration, which is the one that gets forgotten—, updates, video network, accounts and remote access over VPN. It is your own infrastructure and it is administered like the rest of the house.
• Security: the whole operation, which here really is a job in itself. Views, maps, search, bookmarks, exports and attending to the alerts. The operating tool does not let you touch the configuration, so this part can be handed over whole, and without fear, to whoever has to use it.
• The nuance of this platform, and it counts in its favour: because administering and operating are separated, the split can be written down exactly as it is —IT administers, security operates—. In systems where everything is done from the same screen, that split is a promise; here it is a setting.
• The border you have to decide all the same: the rules and the alerts. They are configured on the administration side and they are decided by whoever answers for the risk. You have to sit the two of them round the same table once a year, because the installation changes and the alerts stay as they were.
• Of the integrator it demands the classic craft —sizing, network, storage, integration— and maintenance that really gets done, because there is no cloud supplier watching the state of your system. And knowing the platform: it is less common than the big ones, so whoever installs it has to have installed it before.
• Training at two levels —operation and administration— and the usual documentation plus one thing particular to this platform: the recovery procedure for the configuration server, with its backup and its steps. Plus architecture, retention calculation, a sheet per camera, views and alerts in plain language, and the administrator passwords in your name.

Almost nobody chooses video software by looking at a single platform. These are neighbours in the same conversation, and the index leads to all twenty.