Blade Server Problems
Blade servers are widely used for web hosting and high-density cluster computing. They are essentially stripped-down rackmount servers that are placed vertically into a rack rather than horizontally like the rackmounted servers. The main purpose of a blade server is to reduce space and reduce the number non-essential components. Many blade servers can be hot swapped.
-
Component Failure
-
Unlike tower and rackmounted servers, blade servers rarely have replacement parts that can be bought online. Since most blade servers are bought from individual companies (Dell and IBM, for example) as a whole, the manufacturer has to replace the part for you, which can be a big hassle.
Cooling
-
Because blade servers are packed very close to each other in a rack, they heat up quickly. All around, most server problems, whether they are tower rackmount or blade, are caused by overheating. When you have 20 blade servers, each with one or more redundant power supplies, cooling can be a big problem.
-
Power Consumption
-
Every blade server has one or more power supplies. When you have a blade server farm that consists of multiple racks and runs 24 hours, the building's electrical system can sometimes fail if it is overloaded. It is essential to have backup power sources supplied to each blade server rack and an electrical system that can handle the power consumption.
Replacing a Rackmount with a Blade Server
-
For people who want to replace their rackmount servers with blade servers, many modifications are needed. Mainly, if you had a management system (Ethernet Network or Fiber Channel SAN, for example) that you used for rackmount servers and you want to implement it into your blade servers, you will have to change many different things associated with the setup (networking) of the blade servers.
Failover Rate
-
Blade servers have a high failover rate due to inherent cooling disadvantages, which explains why very many blade servers are made to be hot swappable. Also, the performance of rackmount servers can not be very optimized due to the scarcity of aftermarket parts, which reduces their reliability. For example, Dell has a good blade enclosure but has a motherboard that short-circuits easily. IBM, on the other hand, has a bad blade enclosure but a reliable motherboard. The main problem is that blade servers are made to only be compatible with a specific manufacturer's parts.
-
References
- Photo Credit Image by Flickr.com, courtesy of Steve Jurvetson