Updated: Jan 22
It’s all about the heat
Network uptime on dedicated servers is essential to running an online business but there are other factors that can come into play that will decide whether your dedicated server stays online or not. A lot has to do with the cooling and power at the data center. Hardware failures on dedicated servers can be numerous.
If the data center is not cooling your server enough it will overheat and shorten the life of the hardware. This will be a sure cause of hardware failures on dedicated servers. A lot of data centers have sufficient cooling but there may be instances where not every rack is getting the cooling it requires. In the short term, this may be possible in order to save costs on cooling. However, in the long term, the heat will shorten the life of hardware in that server.
Heat is one of a dedicated server’s worst enemies. It is what causes hardware to eventually fail (along with a few other factors which will be mentioned later.) This can result in expensive hardware replacements down the road or simply having to replace the dedicated servers. Customers, of course, do not want to be on one of the servers that will face a hardware failure but there is, of course, no way for a server administrator to tell unless they have something like IPMI where they can check server temperatures and even be notified of any imminent problems due to high temperatures.
During summer months is when cooling a data center becomes most difficult and most expensive. If a datacenter also does not have backup power this can be a serious problem especially if the location where the data center is located is prone to brownouts. Every reputable data center should have backup power available at all times, no exceptions.
Spikes and surges in electricity can also cause hardware failures on dedicated servers. They are very harmful to hardware in dedicated servers. This is very hard on hardware and will shorten the life of a server’s hardware if surge and spike protectors are not in place. This can easily be answered by simply asking a dedicated server provider if they provide this hardware. Not only will this help their servers last longer, but there should not be any upset customers due to failed hardware that could have been avoided. This does not mean, however, that simply providing cooling and power protection will avoid all hardware failures.
Sometimes hardware will fail due to constant running or powering on and off. This is to be expected and backups should always be in place to prepare for hardware failure.
Let ’em run hot
A growing trend is to also simply let dedicated servers run at higher temperatures. They are simply able to tolerate these higher temperatures. If the manufacturer has an acceptable operating temperature and is operating within that range, then there is no reason why this should be a problem with the life of the hardware. However, it is a tendency for some dedicated server providers to push the limits. As a result, this can save a small fortune in cooling costs, especially if it is a large data center. The question is, will the hardware last as long and how much sooner will hardware or the server need to be replaced at the expense of client-server downtime? Evaluating the cost of hardware failures on dedicated servers is a plausible scenario.
Overclocking CPU’s will force unnecessary strain. This way a dedicated server provider can offer faster CPU speeds without having to pay for the more expensive CPU with the higher speed. Again, the only real way to tell if the CPU has been overclocked is if you have access to IPMI and can check that voltage setting are correct for that particular CPU. Overclocking a CPU can certainly cause stability problems if done incorrectly.
The question is, is it fair for a provider to overclock CPU’s to get more power out of a CPU?