Differentiating between blade, rack and tower servers

October 30th, 2012 by martin237

I got a great email the other day from a reader:

“I came across your blog on google, I’ve just started looking after our first batch of servers, one of the servers that has been delivered is a Dell PowerEdge R720, how do I know what kind of server it is and what’s the difference between the different types, apologies if it’s a stupid question but I’m new to the server space”

I replied quickly and thought I would put together a post on this. Let’s cover some basics first for Dell and HP servers.

Dell server models (PowerEdge)

T servers are their tower range of servers – the T110 II is their entry level tower server for example
R servers are their rack based range of servers, the R720 being their midrange 2 socket rack server
M servers are their blade based range of servers, an example being their M820 blade
C servers are their high density range of servers, an example is their C6100 rack server comprised of four two socket servers

HP server models (Proliant)

ML – their tower range of servers, an example would be their ML110 G7 or the MicroServer (we have both in Bladewatch labs)
DL – their rack mount servers, an example being the DL360 Gen8 servers
BL – their blade based servers, an example being their BL460c Gen8
SL – their scalable or cloud, high density compute range of servers, for example the SL320s Gen8

The underlying technologies between platforms tend to be similar, the same chip that is available in the rack server should be available in the tower, the blade or the high density compute node with a few exceptions. What is different is the dependencies in how the platforms operate, their typical usage scenarios and options available.

Tower servers – often seen as the entry level server offering, they range from a server which is designed for file and print up to ones that can host mulitple applications, have multiple processors, hard drives and memory support. The servers are stand alone and typically have no dependences or linked components.
Rack servers are designed to be hosted in a computer rack which is measured by industry standard dimensions, a single u being approximately 1.75 inches high. These servers are therefore plugged into a rack secured, power, networking and storage are then connected as appropriate. The bigger the server (in u size), typically the more powerful it is, with more capacity for components and upgrades, though in actual performance and energy efficient terms bigger is not always better dependent on the application and the configuration. These became popular with the ‘pizza box’ servers, the DL360 being an example.
Blade servers were released some time ago now in an effort to meet requirements for deploying servers in volume using a modular approach, consolidating for example the network and power into an enclosure. I then plug the blades into this enclosure for network, power and storage.
High density servers – these are a new range of servers with a specific useage scenarios such as hosting companies, high performance computing or cloud type scenarios. They are a hybrid of blade and rack server technology, where I might want to bring some of the benefits of the blade offering but meet specific customer demands like support for GPUs, or have a normal sized rack server which hosts four mini servers inside it.

Posted in Cloud, News, Servers | No comments

Previous post: What will IT want to know in terms of requirements
Next post: What’s an ASR and how do I fix it?

Differentiating between blade, rack and tower servers

Leave a comment