Since first arising in the early 1990s, the x86 server market has all but completely displaced mainframes and minicomputers.
Most of these servers now get deployed in standard 19” rackmount frames within data centers and server closets, although even a small company of 10 or fewer people can benefit from running a standalone server in a tower chassis.
In general, the hardware selected for a server computer will depend on the server application. Different applications, as well as the number of users expected to access the machine, will demand different types of hardware configurations. While there are many kinds of server applications—application, database, file server, Web server, and so on—server hardware is often optimized for a particular mix of three resources: compute capability, storage, and network connectivity.
An obvious example is the Aberdeen Stirling X888 storage server shown below, hosting 48 hot-swappable drive bays in the front plus another two in back. Obviously, so much storage demands a lot of bandwidth on several fronts, which is why the server comes equipped with sockets for two Intel Xeon 5500 Series processors, quad-GbE LAN capability, and an eight-port SAS controller for connecting up to eight additional JBOD storage controllers.
Obviously, this 8U configuration is optimized for high storage density. When it comes to compute resources, though, it’s not feasible to stack processors quite like hard drives. The heat generated by modern processors would be far too excessive for such a container given the amount of ventilation and cooling that can be practically applied. However, design evolution has proven that configurations with one, two, and even four CPUs can operate on a single motherboard within the confines of a single “U” (for 1.75” rack unit).
For many years, 2U servers occupied a key position of popularity in the market, offering a choice balance between processing power, storage capacity, expansion options, and affordability. More recently, though, thanks in part to more energy efficient components and improved board-level integration, that has changed.
“Lots of applications that people used to need 2U for can now be done by a 1U,” notes Charles Liang CEO of Supermicro. “Similarly, our Twin can deliver in half a U what people used to buy in a 1U.”
Until the rise of blade servers in the second half of the last decade, 1U servers were the favored way to maximize compute density in a rack. Today, 1Us remain popular in part because blade systems tend to be quite costly owing to their much more advanced integration. The Supermicro “Twin” concept creates an interesting spin on the 1U model.
William Van Winkle has been a full-time tech writer and author since 1998. He specializes in a wide range of coverage areas, including unified communications, virtualization, Cloud Computing, storage solutions and more. William lives in Hillsboro, Oregon with his wife and 2.4 kids, and—when not scrambling to meet article deadlines—he enjoys reading, travel, and writing fiction.
See here for all of William's Tom's IT Pro articles.
Check Out These IT Videos