WestPoint · Video software · Bosch BVMS
In almost every system the images go from the camera to a server and the server writes them to disc: if that server stops, recording stops. BVMS is built another way, and that decision is the reason for choosing it and the reason it is sized differently.
The video management software from Bosch, meant for installations where not recording for two hours is an incident that has to be explained in writing.
In critical infrastructure, in industry with safety obligations, in transport or on sites subject to inspection, the question is not how many cameras get watched. It is what has to fall over for recording to stop.
The BVMS answer is to take the server out of the way: it hands out space on the storage and the cameras write themselves. If it goes off, you lose the administration and the alarm handling, not the video.
And there is a second safety net: if the network is cut, the camera records on its own card and when it comes back it uploads what was missing by itself. For cameras at the end of a link that drops when it rains, that is exactly what was needed.
A server that configures and directs, a manager that hands out the storage, and the storage, which here is the piece that has ruined most projects.
The central server holds the configuration and manages alarms, users and permissions. The recording manager assigns each camera its space and keeps track of what is recorded and where. Operators and administrators work in different programs.
The storage is block storage, iSCSI type, not a network folder. Sizing it wrong does not show at handover: it shows months later, at full load, in gaps of seconds that come out when all the cameras are writing and somebody is playing back as well.
For several premises there are two schemes and one in two gets chosen wrong. The big one joins complete systems under a higher-level administration. The light one is for small premises that connect only when somebody looks, and it is the right one for forty installations on a domestic line.
The redundancy of the central server protects the administration and the operation: the recording was already protected by the architecture. That distinction lets you spend the money where it is needed rather than where it sounds better.
This is a complete house: video, intruder detection, access, fire and public address are all from the same manufacturer. That saves a lot inside, and you have to look at what happens outside.
The brand's intruder panels come in directly: the zones are items in the system, they are armed and disarmed from the operator's screen and an alarm opens the camera that corresponds to it. This is the good version of «integrated»: with no gateway and no two support desks.
The same happens with its access control, and on large installations there is a higher layer that brings building, fire and video together in a single room. Outside the house the third-party catalogue is shorter: there is an SDK, and the model and version are checked before anything is promised.
Third-party cameras come in over ONVIF and are recorded, but direct writing, gap filling and on-board analytics are functions of this house's cameras. Setting up BVMS with others is paying for an architecture and leaving out the three reasons it gets chosen.
A base licence according to size and camera channels that are added on. Growing means adding channels, which is easy to budget for and to explain to purchasing, which is not nothing.
What is bought separately and disappears from the quotations: joining several premises under one administration, the control room keyboards and some integrations. A multi-site scheme without those lines will show up on the second invoice.
On top of that, the annual maintenance, which entitles you to new versions. On a platform so tied to its own cameras the trap squeezes harder: the cameras you buy in three years' time may ask for a version you can no longer install.
And the cost that is larger here than on the others: the storage. Not because it is dear, because it is demanding. Real sustained writing is needed, and saving on that line is buying yourself the recording gaps of six months' time.
It is the clearest of the six on this, because its virtue and its condition are very specific.
Whoever is going to put in cameras from this brand and needs there to be no gaps: critical infrastructure, industry with written obligations, transport, sites subject to inspection. Whoever already has fire, intruder detection or access from the same manufacturer. And installations spread out with poor links.
Whoever has a large estate of cameras from another brand and is not going to change it: they lose the direct writing, the gap filling and the camera analytics, which are the three reasons for buying. And whoever needs to integrate many different third parties.
In the storage, nearly always. A well-configured system with storage that is just enough works at handover and fails six months later. We ask to see the sustained-write calculation in any quotation, ours included; if it does not exist, it was done by eye.
Here half the value is in two things: the storage and the configuration, which is powerful and does not forgive.
It is comfortable for the operator and demanding on whoever configures it. That asymmetry decides the split.
IT keeps the server, the video network, the storage and the configuration backups. With one critical task: watching the storage, and not only the free space. Here a storage problem does not bring the system down: it leaves silent gaps, which is worse.
Security keeps the operation: users and permissions, alarms and their priority, video exporting and, if there is integrated intruder detection, the arming and disarming of zones. The operator program is reasonable to learn and the security team can run this on their own.
What is not left in the day to day is the configuration program. Changing the retention, moving a camera to another unit or touching how the space is shared out are two clicks with consequences in terabytes. That access is limited in writing and with names, not by custom.
What is left to the integrator is the underlying changes, the growth of the storage, the version jumps and the gap review, which needs knowing where to look. A client operates this very well and almost nobody reconfigures it without help; better to say so before signing.
Training: half a day for the operator at their real position, including the intruder detection if it is integrated, because arming from the video is what gets used most and what frightens people most. A day for the administrator and a session for IT on storage and backups.
Here the analytics are not a module you buy: they are in the camera and they arrive with it. That has one big advantage and one very specific limit.
The house's cameras bring it inside, with no separate licence in most cases: lines that are not to be crossed, forbidden zones, objects that appear, people standing still where they should not, counting. There is no analytics server to size and no fee per analysed camera.
And the camera keeps a description of what happened alongside the video, so you can search afterwards over what was recorded without analysing it again. It is the function that saves most time and the one used least, because nobody taught it.
Number plates, third-party analytics and sector tools can be added, with the usual warning, which weighs a little more here because the third-party catalogue is short.
The real limit is the limit of the whole approach: it is rule-based analytics over geometry and classification. You describe a shape, a crossing or a time and the camera watches that: it works well and it solves most of what gets asked for.
What it cannot do is understand the scene: that a situation is turning bad, that somebody is doing something unsafe that does not consist of crossing a line, that what is on screen is odd even though it follows the rules. For that you would have to describe the world with lines, and it does not have them.
Since this is a recommendation with an interested party in it, we say so before making it: in the same house there is IRIS Neural, the NVMS from Infinity Neural, the other company in the group. It is our own product. If we did not say so, everything above would be worth less.
They live together well: BVMS governs the cameras, the recording without gaps and the control room; IRIS brings the understanding of the scene without touching the recording architecture. And if what is needed is rules, they are already included in your cameras and there is nothing to buy.
If what worries you is not having gaps in the recording, this platform and the rest do not compete on equal terms. If what worries you is being able to choose the camera, they do not either.
Get started
If you have a BVMS quotation on the table, ask for that calculation: cameras per unit, sustained write speed and margin. It is what separates an installation that holds up from one that gives gaps six months later. Tell us how many cameras there are, how many days have to be kept and what else you have from that house, and we will go over it.
