In general, Hardware-level Virtualization provides better isolation and it is considered as a full virtualization platform. Operating System virtualization, however, is often considered as jail-like containerization. In this post, I list a few of top benefits as why you may give OS virtualization technology a try.
Reason 1: Performance
Reason 1: Performance
Performance is the main reason you choose operating system-level virtualization. Unlike Hardware-level virtualization that emulates the entire hardware layer and creates a dedicated kernel on each Virtual Machine, operating system- level virtualization like OpenVZ useschroot environments that rely on the same kernel. The emulation layer created by Hardware-level virtualization has greatly degraded the overall performances (I/O, Network throughput), as well as wasting the CPU cycles and requiring the dedicated memory to run custom kernel on each virtual machine. Generally speaking, the overall performance degradation in Hardware-level virtualization is distinguishable at most scenario, especially in the high-traffic web loads and I/O intensive applications.
Reason 2: Density and Utilization
In general, the operating system-level virtualization has 3 times higher density than its hardware-level counterpart due to the nature of design. Hardware-level virtualization creates an emulation layer called Hypervisor so that its "guest" system would "falsely" treats the underlying hardware resource as the dedicated resources. The Hypervisor further dedicated the resources (such as OS kernel, drivers, user applications and tools) on each "guest" system. With that being said, about 20% of hardware resources have been used for virtualization platform in various processes (known as "overheat" in the virtualization industry).
Operating system-level virtualization, however, uses the single OS kernel in its underlying hardware node. It's Virtual Environment (known as VE) is just a bunch of files and process in that kernel. The underlying kernel manages all the processes across the VEs in the same hardware node. As a result,, OS-level virtualization produces near-zero overheat in virtualization process.
Soon after you come down to the realization of how the virtualization platforms utilize the hardware resources. You have find no trouble why OS-level virtualization has higher "guest" density then hardware-level one.
Reason 3: Agility
Agility is another reason you may think about choosing operating system-level virtualization. To eliminate the misunderstanding, I would like to put some efforts on explaining what "agility" really mean in Virtualization world.
Agility indicates how flexible and easy a virtual instance can be changed in size at any running state of server operations. In other words, how easy we can "scale-up" and "scale-down" the server resources at any time we want it - without interrupting the on-going processes and operations.
In operating system-level virtualization platform, since all of virtual instances are running on the same OS kernel and are sharing the same hardware and OS resources, the resources that are allocated to a specific virtual instance can be changed at the run-time. But, in hardware-level virtualization, since the resources and custom kernels have been provisioned in the course of virtual machine creations, it's almost impossible to dynamically re-size the same instance at the real-time. The common practice of resizing the instances is to spin up a new instance with either larger or smaller resources and then transitions user's files, processes and user configurations onto the new instance - the Linux kernel and visualization platform make this transition possible but require additional techniques involved.
Reason 2: Density and Utilization
In general, the operating system-level virtualization has 3 times higher density than its hardware-level counterpart due to the nature of design. Hardware-level virtualization creates an emulation layer called Hypervisor so that its "guest" system would "falsely" treats the underlying hardware resource as the dedicated resources. The Hypervisor further dedicated the resources (such as OS kernel, drivers, user applications and tools) on each "guest" system. With that being said, about 20% of hardware resources have been used for virtualization platform in various processes (known as "overheat" in the virtualization industry).
Operating system-level virtualization, however, uses the single OS kernel in its underlying hardware node. It's Virtual Environment (known as VE) is just a bunch of files and process in that kernel. The underlying kernel manages all the processes across the VEs in the same hardware node. As a result,, OS-level virtualization produces near-zero overheat in virtualization process.
Soon after you come down to the realization of how the virtualization platforms utilize the hardware resources. You have find no trouble why OS-level virtualization has higher "guest" density then hardware-level one.
Reason 3: Agility
Agility indicates how flexible and easy a virtual instance can be changed in size at any running state of server operations. In other words, how easy we can "scale-up" and "scale-down" the server resources at any time we want it - without interrupting the on-going processes and operations.
In operating system-level virtualization platform, since all of virtual instances are running on the same OS kernel and are sharing the same hardware and OS resources, the resources that are allocated to a specific virtual instance can be changed at the run-time. But, in hardware-level virtualization, since the resources and custom kernels have been provisioned in the course of virtual machine creations, it's almost impossible to dynamically re-size the same instance at the real-time. The common practice of resizing the instances is to spin up a new instance with either larger or smaller resources and then transitions user's files, processes and user configurations onto the new instance - the Linux kernel and visualization platform make this transition possible but require additional techniques involved.





