The title is telling, I guess. In particular, I am wondering how risky is it not to change the fan of a 4 years old Supermicro server?
The life expectancy of a server grade fan
fanhardwaresupermicro
Related Solutions
As you suspected, small fans and high RPM make for a high noise level. These fans aren't good at anything except making enough pressure to get air past obstructions. In choosing your cooling mod tactic, take a good look at the system and see if the tiny fan is blowing air on one particular component, or just drawing air out of the case in general.
Replacing the fan with a large, slow one is definitely an option, providing you don't mind voiding the warranty for the modifications required. In doing the change, keep in mind that slower fans don't build as much pressure, so you have to provide larger, more free flowing passageways. That means you'll probably need to remove some metal, make a large hole or many small ones, to increase the airflow. If you're blowing in, you'll need to carefully seal the edges of the fan against the case. Silicon hobby glue or gasket maker is good for this.
If you're drawing air out, you should use the natural convection tendencies of the system by putting the fan out top, blowing upward.
In the days of 80mm case fans, I modded every system case I bought to a.) provide as many fans in as out, and b.) replace as many 80mm fans with 120mm fans as possible, and c.) use lower RPM fans, and more of them. The result was always cooler and quieter then the original setup, even if the original had only two fans.
To liken it to a car analogy: A taxi can do over 500,000 kilometers before it needs an engine rebuild. The reason for this is because they are always running, 24/7, and after a car's engine is up to temperature, the amount of wear it receives while it is running is greatly reduced.
A computer is kinda the same. The majority of the "wear" on parts can happen when the server is booting up. Just attach an amp meter to your computer, and turn it on. When it starts up, the power it draws climbs very high, and then it settles down once all the disks have spun up and the processor is initalised. Also, think about how much disk activity the server undergoes during boot up vs when it's working. Chances are the disk access from booting the OS is fairly solid activity, whereas when the OS is running, unless it's a very heavy database server (I'm guessing not), the disks will most likely stay fairly idle. If there's any time it's going to fail, chances are it will be on boot up.
Turning your server on and off is a stupid idea. Not only to mention most servers can take upwards of 2-5 minutes to just get past the BIOS checks, it's a huge amount of wasted time too.
2018 Update: Given that most computers are now essentailly entirely solid-state, this answer may no longer be as accurate as it once was. The taxi analogy doesn't really suit todays modern servers. That said, typically you still generall don't turn servers off.
Best Answer
This is an anecdote, not evidence. I've had (SuperMicro) server fans in service for 7+ years of continuous spinning without a problem.
I've also had one utterly catastrophic failure of one after five years. It became discolored and ultimately one of the blades shattered. We believe that it was the root cause of a multiple-drive failure in that machine. I keep it on my desk as a reminder to always make sure that hardware monitoring is set up, active, and properly configured to send alerts.
You can ask the server itself for information on how fast the fans are spinning. This information is available from the on-board sensors. Under Linux, look into
lm_sensors. If your servers have IPMI, you should be able to get that information out of any IPMI tool as well. A fan that is acting up or that has failed will unexpectedly have a low or zero RPM.