Guide
On-premise vs cloud video analytics: which fits your site?
On-premise analytics runs on hardware at your site, so video stays on your own network. Cloud analytics sends video or clips to a provider's servers, so there is less to install but more depends on the internet and on the provider. Large or regulated sites usually choose on-premise; homes and small shops often find cloud simpler.
5 min read
Where the analysis runs
Both approaches do the same job: software watches camera video, spots events and raises alerts. The difference is where that software runs. With on-premise analytics, it runs on a computer at your site, next to your recorder, and the video never has to leave your network to be analysed. With cloud analytics, video travels over the internet to the provider's servers, is analysed there, and the results come back to you. That video may be a continuous stream or short clips around an event. Many products sit somewhere in between, so ask each supplier what runs where.
Where the footage goes in each
On-premise
Video goes from the cameras to your recorder and to the analytics hardware, all inside your network. What leaves the site is usually small: licence checks, software updates and alerts to people's phones. You decide who can see footage, and the record of who looked at it can stay on site too.
Cloud
Some video has to leave the site, because that is where the analysis happens. Providers handle it in very different ways:
- Continuous streaming: every frame from the chosen cameras is uploaded.
- Event clips: only short clips around possible events are sent.
- Storage: some providers keep clips or recordings for days or months; others delete clips once analysed and keep only a text record.
Questions to ask any cloud provider:
- What is sent, and when?
- Is it encrypted on the way?
- Where is it processed and stored, and in which country?
- How long is anything kept, and can you delete it?
- Who at the provider can see it, and is that access logged?
Internet dependence
On-premise
Detection and on-site alerts carry on if the internet connection drops, because everything needed is local. Notifications to phones away from the site wait until the connection returns. Viewing cameras remotely needs the internet, as it always has.
Cloud
Analysis stops when the connection stops. Your recorder keeps recording, so nothing is lost from the recording itself, but events during the outage may be flagged late or not at all.
Upload capacity matters too. Many connections have fast download speeds and much slower upload speeds. Uploading several high-resolution cameras continuously needs a lot of upload capacity; sending short event clips needs far less. Check your connection's upload speed as well as the advertised download speed.
Cost shape: hardware vs usage
Comparing costs is tricky because the two models spread the money differently.
On-premise
On-premise usually means buying computing hardware sized to your camera count and the features you want, plus a software licence. The hardware is a one-off cost that you own; the licence is often yearly. Costs grow in steps: adding many more cameras may eventually need more hardware.
Cloud
Cloud usually means little or no hardware and a subscription, often per camera per month, sometimes with charges for usage or storage. It is easy to start small. Over several years and across many cameras, the running total can overtake the hardware cost of an on-premise system, and you also pay for the internet capacity it needs.
A fair comparison uses the same cameras and features over three to five years. Include hardware, licences or subscriptions, internet, installation and support.
Which sectors should stay on-premise
Some organisations have strong reasons to keep video analysis on their own premises.
Banks and NBFCs
Financial institutions work under close regulatory oversight of how they handle data and outsourced technology. Many prefer to keep CCTV video inside their own premises, and head offices often want event records from branches rather than video.
Hospitals
Cameras in and around patient areas capture sensitive moments. Keeping that video on the hospital's own systems is the simplest way to limit who can see it.
Sites with data-localisation or contractual rules
Some organisations are bound by sector rules or customer contracts that require data to stay in India, or on their own premises. Check these before you choose a cloud service.
Privacy law applies either way. India's Digital Personal Data Protection Act, 2023 calls the organisation that decides why and how personal data is processed the Data Fiduciary, and a company that processes it on that organisation's behalf a Data Processor. With most CCTV analytics, you remain the Data Fiduciary for your footage, whichever model you choose. This is a general explanation, not legal advice; take advice on your own situation.
A simple way to decide
Answer these in order:
- Do rules, contracts or company policy require video to stay on your premises? If yes, choose on-premise.
- How many cameras do you want analysed? Dozens or hundreds usually favour on-premise; a handful often suits cloud.
- How reliable is your internet, and how much upload capacity do you have? Weak or shared connections favour on-premise.
- Do you have somewhere to put a small server, and someone to keep it powered and connected? If not, cloud is simpler.
- What does each option cost over three to five years, counting every year after the first?
If most answers point the same way, that is your answer. If they are split, run a short trial of each on your own cameras.
Where SurveyLensys fits
SurveyLensys runs on-premise for larger and regulated sites: detection and alerts run on hardware at your site, and only licence checks and software updates connect to us. For homes and small sites, a cloud option is in early access: short event clips are sent over an encrypted connection, analysed and deleted straight away, and only a text record of each event is kept, plus an optional small thumbnail. To see which suits your site, book a live demo on your own cameras.