Depends on a lot of things such as how Hot the Environment it is kept in is, how many times it has a sudden power Shutdown rather than a Graceful Shutdown.
Heat is an important factor. How it shuts down really isn't. With a rude shutdown the actuator moves to a "neutral corner" and the platters stop spinning. Nothing bad happens mechanically. THe actuator does not dive into the platters and dig a trench in the recording media.
The risk from a rude shut down is if the drive is in write mode and a few bytes get sprayed across sectors they were not originally targeted for. This can corrupt a file or two on occasion, but unless it is a file important to operation, such as the OS, there is no real danger, and even if those files are important it only means a loss of data. If it sprays across the catalog database you could lose a lot of recordings, so it is not highly recommended to yank the power cord, but will it hurt the drive mechanically or electrically? Categorically, no.
There are now hard drives designed for use in DVRs that run 24/7. I would say most drives last around 5 years but could go on for 7 or 8 years depending on the factors stated above and other factors such as dependable noise free electricity.
The jury is still out on whether a drive has a shorter life expectancy if it regularly shuts down, or if it continues to spin. Either scenario has the potential for mechanical failure; for one, the bearings wear out faster, for the other the inertia of spinup and spindown may be even worse. No one really knows.
"Noise-free electricity" most certainly has nothing to do with it; AC is reduced to DC in every power supply, which means all spikes or frequencies much above zero Hz are completely filtered out. When noise becomes a problem is when the filter caps dry out letting ripple and other noise through, but it will play hell with your microprocessor long before your HDD will ever notice it. The early warning sign that your PS may be letting noise into the system is that the device reboots regularly. IOW, you have larger problems long before things can deteriorate to the level of threatening the chips in the HBA or HDD.
The most definitive information is in a Google white paper (which you can google--I did) which did a comprehensive study of thousands of HDDs over an 8-year period in an enterprise environment where they are running all the time. The original report second-hand was that HDDs, after a few-month infant mortality period where many die, fail at a rate of about 8% per year, compounded. That means a HDD on its 2nd birthday will have a 16% chance of failure within the coming year, and that a HDD on its 3rd birthday will have a 24% chance of failure in the coming year, and so on, which means that once a drive gets beyond 5 years old, it has about a 50-50 chance of living to its 6th birthday, and if it makes it to 8 years, it has about a 25% chance of making it to 9 years old. It also means that a new drive has much less of a chance of ever making it that far.
But if you read the white paper carefully (again, I did), the results are not as clear, and are not quite as easy to pin down. Google seems to say that a HDD will have a "significant chance" of failing between 3-5 years. But what is "significant"? 10%? 30%?
Regardless, this has no impact or preventative replacement, which seems to make little sense in most applications and even less sense in a DVR application. The only similar scenario that might make sense is retiring a 3 to 4 year-old HDD, say, in September, just before the new fall season begins. But that entails swapping drives if you want to go back (a Thermaltake helps if that is what is decided).
Also, who made the drive and what type of drive is it.
According to the Google study, really not so much. Every drive manufacturer (and I think there are really only 3 left) has had a run of problematic drives on occasion, but generally they are all about the same as far as shelf life. HDDS don't have "types". They run at different rotational speeds (in a DVR slower is actually better because of the heat factor), they have different chips in the controllers, they have different capacities, buffer sizes and physical sizes, but they all work in pretty much exactly the same way. Those designed for media applications don't do thermal calibration the same way, but those not designed for media will last just as long; its just that they will never perform well even out of the box in a DVR scenario.
The Power Supply will probably go first on most DVRs which Mimics a Bad Hard Drive.
Actually, the HDD will probably has a life expectancy twice the normal HDD. It has no moving parts, for one thing. HDDs wear out over time; either the actuator sticks or dies, a bearing sticks, the recording surface flakes off, or there is a catastrophic failure of an electronic component. PS's degrade slowly, and then eventually die from a power surge or catastrophic failure of the voltage regulator. But the average life of a PS is still longer by far; its a solid state device while a HDD is mainly a mechanical device. But you are correct in that a degrading PS with drying filter caps will often present very much in the same way (spontaneous hangs and reboots) as a dying HDD.
Actually Richie, I Think The Worst Threat To a HDD is Annoyingly-Improper Capitalization.