Two platforms under the same name. And they are not the same.

Avigilon —today part of Motorola Solutions— became known for cameras, not for doors. The access control came later and by two routes that have never merged into one product. That is why the brand name, on its own, does not tell you what you are going to install.

Unity Access is today's name for what was called Access Control Manager: a server on site, meant for large offices. Alta Access is what was bought as Openpath: a cloud service, with no server in the building, with the phone as the normal way of opening.

Both solve the same old thing —who opens which door, when, and that it is written down— from opposite ends. They turn up where the video is already from this firm: a head office, a campus, a network of offices. The conversation does not start with the doors, it starts with «we want one single screen».

Unity is classic architecture: a server inside, the database there, operators working against it. But what counts is not in the software, it is in the cabinet: the door electronics are from the Mercury family, from HID, the same one several enterprise platforms from other brands use.

That has a consequence almost nobody asks about before signing and it is the one that saves the most money three years later: the panel is not captive to the software. The day the head has to change, the cabling, the readers and often the controller stay where they are.

Alta has no server. Each controller talks to the tenant in the cloud and keeps its copy of permissions and schedules, so it decides locally. Unity grows inwards —partitions, interlocks, anti-passback by zone, evacuation lists—; Alta grows outwards: thirty branches in the same tenant and no server to maintain. The two hardly ever both fit.

The people, the permissions, the schedules and the log live in the cloud. The decision lives in the building: each controller has its copy and opens without asking anybody. If the line goes down, the doors carry on with the last configuration and the events are stored to be uploaded later.

What you lose with the line down, specifically: a change is not propagated —you remove somebody and their credential goes on opening until the controller finds out—, you cannot open a door remotely and you cannot watch it live. The phone credential, on the other hand, does not need the internet: the phone and the reader understand each other on their own.

And the data, which is not a formality at the end. The manufacturer is American and the one who answers to the Spanish data protection authority is the client. Three things are asked for in writing: which region the data is hosted in, the data processor agreement with its sub-processors, and what mechanism covers transfers outside the European Union.

Unity fits where there is a large head office or a campus, a security department with an operator and specification requirements: zones, interlocks, a count of the people inside, auditing. And it fits above all if the video is already from this firm, because then the single console stops being an argument and is a fact.

Alta fits where there are many small sites and none of them has an IT person: scattered offices, clinics, shops, industrial units belonging to the same company. Also where you want the server and its maintenance out of the way.

Neither fits if there is a door and a half and nobody inside to handle the joiners and leavers. Alta is out if the company policy is that no personnel data leaves its own servers, and better the conversation ends there than three meetings later. Unity is out if nobody is going to maintain that server: a security server that is never updated ends up being the biggest hole in the building.

Adding and removing people, the schedules, the permissions by zone and the groups can be handled by the client's team, security or IT, and they should be. Both lines let you give somebody the management of people without giving them the configuration of the system: whoever registers a worker has no need to be able to change what a door does.

What the client does not touch is the layer underneath. Adding a controller, changing a door's mode, touching the integration with video or intruder detection, rebuilding the zone tree or moving up a version. It is not about keeping the work: it is that a mistake there is not seen until the day the door is needed.

For them to really be able to, two things are handed over. Training by role —whoever adds and removes people, whoever looks at the log and exports it, whoever deals with a door that will not open— and documentation that is of use: a door map with its behaviour, the zone and schedule tree exactly as it ended up, who holds each role, and one sheet with the five problems that are going to happen.

If what you have is an enterprise platform on Mercury controllers, migrating to Unity is about as painless as it gets: the cabling, the readers and often the panel stay. It is not free, but no walls have to be opened.

The history is not kept. It is exported to a file and stored separately before the old system is switched off: it is not going into the new platform, and there are sectors where somebody will ask for it two years from now.

And the real stumbling block is the cards. The old proximity ones can usually still be read, and this is the moment to stop doing so because copying them costs next to nothing. The ones encrypted with the previous manufacturer's keys normally cannot be read, and then there is a reissue for the whole workforce: that is a morning of queueing at reception, and it gets planned.