A server farm or server cluster can be a range of computer servers – usually maintained by a business to supply server functionality far beyond the power of a person machine. Server farms often include a large numbers of computers which need a great deal of capacity to run also to maintain cool. At the optimum performance level, a server farm offers enormous costs (both financial and environmental) linked to it. Server farms frequently possess backup servers, that may dominate the function of primary servers in the event of a primary-server failure. Server farms are often collocated with the network switches and/or routers which enable communication in the middle of your various regions of the cluster and the users of the cluster. Server farmers typically mount the computers, routers, power supplies, and related electronics on 19-inch racks in a server room or data center.
Source: gpuhub render theo gio
Server farms are usually used for cluster computing. Many modern supercomputers comprise giant server farms of high-speed processors connected by either Gigabit Ethernet or custom interconnects such as Infiniband or Myrinet. Web hosting is generally a common using a server farm; such something may also be collectively referred to as a web farm. Other uses of server farms include scientific simulations (such as computational fluid dynamics) and the rendering of 3D computer generated imagery (see render farm).
Server farms are increasingly being used instead of or furthermore to mainframe computers by large enterprises, although server farms will not yet reach the same reliability levels as mainframes. Because of the sheer number of computers in large server farms, the failure of a person machine happens to be a commonplace event, and the management of large server farms must consider this into consideration giving support for redundancy, automatic failover, and rapid reconfiguration of the server cluster.
The performance of the largest server farms (a lot of processors or even more) is normally linked with the performance of the knowledge center’s cooling systems and the total electricity cost rather than by the performance of the processors. Computers in server farms run 24/7 and consume large sums of electricity, and as a result of this, the critical design parameter for both large and continuous systems is often performance per watt rather than cost of peak performance or (peak performance / (unit * initial cost)). Also, for high availability systems which has to perform 24/7 (unlike supercomputers which may be power-cycled to demand, and likewise tend to run at higher utilizations), there may be more attention added to power saving features such as variable clock-speed and the ability to pull the plug on both computer parts, processor parts, and entire computers (WoL and virtualization) according to demand without decreasing services. The network connecting the servers in a server farm is definitely an essential element in the entire performance particularly if operating applications that process massive volumes of data.
Performance per watt
The EEMBC EnergyBench, SPECpower, and the Transaction Processing Performance Council TPC-Energy are benchmarks designed to predict performance per watt in a server farm. The energy employed by each rack of gear could possibly be measured at the energy distribution unit. Some servers include power tracking hardware which means people working the server farm can measure the power employed by each server. The energy utilized by the entire server farm could possibly be reported with regards to power usage effectiveness or data center infrastructure efficiency.
According for a few estimates, for every and every 100 watts assigned to operating the servers, roughly another 50 watts must cool them. Due to this, the siting of a Server Farm is usually often as important as processor selection in achieving power efficiency. Iceland, which includes a cold climate all year and in addition cheap and carbon-neutral geothermal electricity supply, is obviously building its 1st major server farm hosting site. Fibre optic cables are being laid from Iceland to AMERICA and Europe to enable companies there to discover their servers in Iceland. Other countries with favorable circumstances, such as Canada, Finland, Sweden and Switzerland, want to attract cloud computing data centers. In these countries, heat from the servers could possibly be cheaply vented or used to help heat buildings, thus reducing the energy consumption of conventional heaters.