Before starting our review, it’s worthwhile to understand what a hypervisor is and what functions it performs. When performing a large number of various tasks, it is very convenient to use virtual machines. This is due to the fact that they make it possible to host several operating systems on one physical server, each of which is equipped with its own software for solving various problems. Virtual private server has become an excellent option for different types of business. You can find such hosting at https://justvps.com/. Hypervisors are used to provide easy interaction with such virtual machines. Is is software that allows you to manage virtual machines: install, enable and disable them. One of the most popular hypervisors for now is KVM.
It enables virtualization on servers running Linux. A definite plus is that this hypervisor is part of the Linux kernel, so it is constantly being improved and updated. You can use it only in the case of hardware virtualization – using Intel or Amd processors. The KVM processor module allows it to access directly to the core. Thanks to this, you can directly manage virtual machine files and disk images. Each VM has an individual space.
First developed by Cambridge students, this project quickly became commercial due to its promise. Cross-platform and wide functionality of Xen makes its capabilities extensive enough for use in the offices of large companies and corporations. Its core has a paravirtualization mode, that is, it can be configured to simultaneously interact with the hypervisor.
The code for this hypervisor is not overloaded with unnecessary functions. It is equipped with RAM management capabilities, the frequency of the processor, work with direct access to memory, as well as a timer. All other functions are performed by the currently connected VMs. The listed advantages make working with Xen simple and convenient even for a person whose knowledge is not very deep in this area.
In order to compare two hypervisors, you need to test them. For this, two servers with absolutely identical hardware and software were taken, with the exception of, of course, the hypervisors in question. All settings for the created virtual machines were set by default in accordance with the basic settings of both hypervisors. Each option was allocated the same amount of memory.
Let us clarify that the opinions that the distribution kit for the OS we chose – Fedora 20 from Red Hat – are better suited for KVM are not entirely true. We do not consider the struggle of VMs for processor resources during simultaneous operation, since hypervisors can show different performance with varying degrees of competition. Therefore, we consider the competition conditions fair for both parties.
Testing was based on measuring test results for VMs located on hardware alone, excluding software. The deviation in the performance of two servers without virtualization turned out to be less than half a percent.
The first candidate, KVM, showed an average productivity one and a half percent lower than that of iron. In the 7-ZIP test, this hypervisor turned out to be almost 3 percent slower, and in the PostMark test by as much as 4 percent faster than iron. The results of Xen turned out to be worse, he did not surpass his competitor in any of the tests, 2.5 percent behind the iron in three tests, and in the rest he showed himself even worse. The strongest deviation was revealed, as in KVM – in the PostMark test, but Xen did not overtake the iron as its competitor, but lagged by almost 15 percent. Repeated test results deviated from the previous no more than 2 percent.