NVR vs cloud camera storage for a large home

For a large home in Leawood — or anywhere in Kansas City, Missouri and the Johnson County suburbs — the answer is an on-site NVR, for three reasons: no monthly cloud subscription, which saves real money and compounds fast on an eight- or sixteen-camera system; higher recording quality and frame rate, because you are not throttled by an upload pipe; and zero footage lost when the internet drops. A cloud-only camera stops recording the moment the connection does, which is exactly when you would want it recording. Cloud-only systems are fine for a single doorbell. For 4K coverage of a large home, the recorder belongs in the house. I am Daniel Alon, and IntegrateIT installs these systems across the Kansas City metro from Overland Park, KS. Call (913) 804-7575.

That is the short version, and our Kansas City security system cost guide carries it alongside the pricing and stays the canonical reference for both. This is the long version — the arithmetic behind “how much storage,” what the cloud actually asks of your internet, why recorders want a particular kind of hard drive, the one real argument for the cloud that deserves a straight answer, and the hybrid setup that is the right call for most large houses and appears on neither page.

The storage arithmetic nobody shows you

“How many terabytes do I need?” gets answered by salespeople with a shrug and a number. It is arithmetic, it takes thirty seconds, and knowing it changes what you buy. Take the camera’s bitrate in megabits per second, divide by eight for megabytes per second, multiply by 86,400 for a day, then by the number of cameras and the days of retention you want.

terabytes = (Mbps ÷ 8) × 86,400 × cameras × days ÷ 1,000,000

A 4 Mbps H.265 stream — a reasonable figure for a well-configured 4K residential camera, though you should substitute the number from your own camera’s spec sheet — works out to roughly 43 GB per camera per day. From there:

Two things move those numbers more than anything else. Motion-only recording can cut them substantially — but not on a camera watching a street or a driveway that a road runs past, where “motion” is close to continuous, and those are usually the cameras you most want a full record from. And bitrate is a bigger lever than resolution: pushing a camera to a higher-detail stream raises the storage in direct proportion, which is why we set bitrate per camera against what that camera is actually for rather than turning everything up.

The practical takeaway is that retention is the cheapest thing to over-buy at install and the most irritating to add later, because adding it means opening a recorder that is in service. Thirty days is the residential default we build to. If you travel for long stretches, sixty is worth the extra drive, and the arithmetic above tells you exactly what that costs before anybody quotes it.

Why a recorder wants particular drives

A recorder’s workload has almost nothing in common with a computer’s. It writes continuously, sequentially, from several streams at once, twenty-four hours a day, for years, and it never gets an idle night. Surveillance-rated drives are built for that: a stated annual workload rating in terabytes written, and firmware that prioritizes continuing to write over pausing to retry a marginal sector — because in a recorder, a paused write is footage that does not exist.

Desktop drives dropped into a recorder to save eighty dollars are the single most common cause of the call we take most often: live video working perfectly, and nothing recorded for months. When the overwrite cycle fails, most recorders keep streaming live while writing nothing, and nobody notices until the day they need the footage. It is the failure mode behind a good share of our security system takeover work.

Three habits follow from that. Treat the drives as a consumable with a service life rather than a permanent part of the box. Turn on the recorder’s own disk-health alerting, and make sure it emails somebody who reads it. And once a quarter, pull footage from a date a week ago — not from live view — because that is the only test that actually proves the thing is doing its job.

What the cloud asks of your internet

The subscription is the cost people notice. The bandwidth is the cost that decides whether the system is any good. A cloud camera has to push video out continuously, and eight cameras at even 2 Mbps each is 16 Mbps of sustained upload, all day, every day. Residential plans in this metro are strongly asymmetric — download is generous, upload is a fraction of it — so that number lands squarely in the part of the connection you have least of.

Two things follow, and both are bad. The cameras compress harder and drop frame rate to fit the pipe, which is precisely the quality you were buying 4K sensors to get. And the upload competes with everything else the household does — video calls, cloud backups, gaming, a spouse working from home — so the cameras quietly degrade the internet and the internet quietly degrades the cameras. A local recorder moves that entire conversation onto your own switch, where the bandwidth is effectively free. Cameras are PoE devices on switch ports either way, which is why the switch and the cabling are part of a camera system whether or not the quote itemises them — the network build and the camera system are one project pretending to be two.

The one real argument for the cloud

Somebody takes the recorder. It is the honest counter to everything above, and dismissing it would be dishonest: a box in an unlocked closet by the back door is a single object whose removal takes all of your evidence with it. That is a genuine failure mode and it has happened to people.

It also has three good mitigations. Put the recorder where it is neither obvious nor easy — a rack in a mechanical room that locks, not a shelf in the mudroom. Push clips of alarm and analytic events off-site, so incident footage exists in a second place even if the box walks. And make sure the cameras covering the room the recorder is in are among the ones doing that pushing, because the footage of somebody taking the recorder is the footage you most want to survive.

Do those three and the cloud’s remaining advantage is small. Skip them and it is real. We would rather you knew which one you had chosen.

The hybrid, which is what most large houses should actually buy

Record everything locally at full rate. Push event clips off-site. That is the whole recommendation, and it takes the good half of both architectures.

You keep the continuous full-quality record, at the frame rate and bitrate the sensors can actually deliver. You keep recording through an ISP outage, a cut line, or the suspiciously well-timed internet failure that precedes some incidents. You avoid a per-camera monthly bill on the bulk of the data — a real, recurring cost that adds up fast on an eight-camera system and never appears on an NVR quote. And the footage that matters most exists somewhere a thief in your mechanical room cannot reach. The off-site copy stays cheap and stays inside your upload budget precisely because it is clips rather than continuous video.

Two more things belong in the same conversation, because they are what actually keeps a recorder recording. It wants to be on a UPS, so a brief outage ends with a clean shutdown and an orderly power-up rather than a corrupted volume. And it wants somewhere with airflow — recorders are hot, always-on devices, and a cabinet with no ventilation shortens drive life measurably.

When cloud is simply the right answer

A single doorbell. One camera over a detached garage. A rental where you are not running cable. A property with nowhere to put a recorder or cool it. A small business with no IT closet and nobody to look after one. In every one of those, cloud storage is the better product and a recorder would be an expensive way to feel serious.

The line falls roughly where the camera count and the resolution make the per-camera subscription and the sustained upload start to bite — for most households, somewhere around the third or fourth 4K camera. Below it, buy the convenience without apology. Above it, the recorder pays for itself in subscription avoided within a couple of years, and pays for itself again the first time the internet is down during something you needed to see.

How we specify it

We use 4K Ubiquiti or Hanwha cameras and an on-site NVR on every residential install, with the feeds integrated into Control4 or Crestron so cameras appear on the same panel as lighting, audio and climate — and so a driveway camera detecting motion at night can bring the entry-walk lights up. Storage is sized from the arithmetic above against the retention you actually want, not from whatever the recorder shipped with.

What drives the number is camera count, resolution tier, and how large the recorder and its storage need to be for your retention target. Optional professional monitoring is priced separately and billed annually. All of it is installed under IntegrateIT’s own licensed alarm contractor status, so one licensed party permits, installs, certifies and monitors the system rather than handing the alarm to a national vendor. We give one fixed price after a no-cost site walk — not a range by email. The full breakdown of what moves the number is in the security system cost guide.

— Daniel Alon, founder, IntegrateIT. Overland Park, KS. September 2026.