Another league, another budget, and another conversation altogether.

It is continuity. An array of this class has two controllers: if one fails, the other carries on serving the data, and an update of the internal software is done in halves without stopping the service. In mid-range storage there is only one, and that means any fault or update stops the recording for as long as it lasts. For a system that cannot stop even for half an hour, that is the whole difference.

The second difference is data management. These arrays know how to take snapshots of the state of the volume, replicate it to another array in another building and move what is no longer touched to cheaper storage. That is what lets you build something that IS a backup, unlike a set of drives with redundancy, which only protects you from one drive breaking.

And the third is that it comes with a service around it: contracts with written terms, support that watches the array from outside and warns about parts before they fail. You pay for that, and in a critical system you pay for it gladly.

Where the video is critical and the installation is large: infrastructure, airports, big campuses, banking, industry with its own data centre. Places where stopping the recording for half an hour to update a storage system is not a nuisance, it is a breach.

And where the video has to leave its own site. If the policy says the recording from one site also has to exist at another, replication between arrays does it in an orderly, monitored way, which is a good deal better than a home-made process copying files overnight.

And, very often, it fits because it is already there. When the client has a storage team with their array, their contract and their backup procedure, the conversation is not what to buy: it is whether the video can live there, how much room it is allocated and who answers for it.

Video writes at all hours, at a constant rate and almost always in sequence. What that array usually has inside it is the opposite: databases and virtual machines making many small, scattered requests. The two live together well if the bit rate has been counted; if it has not been counted, the video can disturb what was already there or the other way round. You measure beforehand, in sustained megabits per second, and you agree it with whoever administers the array.

Then, the path. If the recording travels over the company network to the storage, that path has to be sized as a constant rate and not as bursts, and you have to decide what happens if it is cut: the recorder needs enough local drive space to carry on recording on its own for a few hours and transfer afterwards. Without that, a network break is a gap in the recording.

And the money, which here is a real conversation: a terabyte in an array of this class costs several times what it costs in mid-range storage. That is why it almost never makes sense to put thirty days of every camera in there. What makes sense is to split it: what is critical and what has to be kept, in the array; the rest, in cheaper storage near the recorder.

In a medium-sized installation, no. Forty or sixty cameras at one site are solved with storage of another class, and doing it that way is not settling for less: it is not paying for continuity and data management nobody is going to use. An enterprise array with no team to operate it is money asleep.

Nor as the single place to put all the video from all the cameras for the whole period. That sum almost never works out, and when it does it is because somebody has trimmed the days of recording to make the budget fit. Better to tell the truth: what matters here, the rest over there.

And we would not fit it without somebody responsible. If the decision is to take the video to the corporate storage, there has to be somebody on that side who says yes, with an allocation, monitoring and a place in the recovery plan. A dependency with no owner is worse than not having it.

An array of this class is operated by the department that already operates the storage in the house. It would make no sense for a security company to administer the corporate array, and we do not claim to. What is needed is for the video to go in there under the same rules as the rest: its allocation, its monitoring, its alerts and its place in the recovery plan.

The boundary we write down is this: from the camera to the recorder, and from the recorder to the place where the data is left, us. From there inwards —capacity, volume snapshots, replication, archiving— the client's team. And a joint review once a year, because cameras multiply and nobody goes back to look at the retention period that was written down on day one.

One thing is not delegated even though the data lives in IT: who can pull a recording out and with what record kept. The video being in the corporate array does not mean anybody in IT can see it. That is decided separately, written down and can be audited.