The X86 compatible hardware of today, regardless of processor count or core count, was designed to run a single operating system. This leaves most machines vastly underutilized. VMware virtualization lets you run multiple virtual machines on a single physical machine, with each virtual machine sharing the resources of that one physical computer across multiple environments. Different virtual machines can run different operating systems and multiple applications on the same physical computer. WMware is an operating system that sits directly on the hardware and is the interface between the hardware and the various operating system. It expands the hardware, from the users point of view, to many different independent servers all with their own processors and memory. These virtual servers cannot be distinguished from physical servers by the end users.

VMware works by loading a small, efficient operating system, or hypervisor directly on the host hardware. The VMware hypervisor has a small footprint and is extremely efficient, with a very small (1%) overhead. Device drivers for nearly all major brand devices are available from VMware. These are loaded during the configuration process.

The client operating systems, such as Microsoft® Server 2008, Microsoft® Small Busines Server, Linux varieties, etc. are then set up as virtual machines, working directly with the VMware layer rather than with the actual hardware. This allows replacement of hardware to be very simple. If the hardware is replaced, VMware is reconfigured for the new hardware, and the virtual guest operating systems see no change whatsoever and are immediately able to boot and operate.

Processors

To explain how processor resources are allocated, first of all one needs to understand how WMware treates processors.

As an example, consider our model 951 HPC Server, with 128GB of memory and Dual 8-Core 2.7Ghz Intel® Xeon™ processors. This would be presented to VMware as a processor pool at ( 16 * 2.7) = 43Ghz. During the configuration process for virtual operating systems guests, each virtual can be assigned a portion of that available processor pool, 22Ghz, depending upon the requirements of that guest. This assignment is most often left to VMware to perform dynamically during operation. Processor multi-threading is ignored by VMware, other than the small gain that will be realized by multi-threading.

Memory

Memory usage can also either be set during configuration or left to VMware to configure dynamically. You may wish to limit memory usage on some servers, such as exchange, that will atttempt to consume as much memory as is available.

Data Storage and Drives

JNCS highly recommends separating data storage needs from the physical server, although VMware is capable of using a drive pool located physcially within a server. Regardless of how the storage is physically accomplished, client oeprating systems are set at configuration to use a portion of the drive pool for their 'boot', or 'C' drives. Other storage can be set up logically so as to available for access by any of the virtuals machines. Again, to users of those machines, the drives may set to appear as physical drives attached to the virtual machine.

Any JNCS server is capable of acting as a phyical VMware host. The speed and core count of the processors, or, as defined above, the processor pool, should be matched to the sum of requirements of the virtual operating systems that will be installed. The required memory capacity is also a function of the requirements of the virtual clients. JNCS will be glad to help in designing a proper configuration.

We strongly suggest using a separate storage server for your storage needs. We configure storage solutions based on Nexenta. If required, we can install storage within the physical VMware host.

VMware allows the enterprise to replace many disparate, underused devices with a few virtual hosts. This greatly reduces system downtime, allows for simple movement of virtual clients from one hardware host to another and allows for scheduled hardware repair or replacement with downtime by moving those clients to another hardware host on the cluster. It also allows the IT administrator to very quickly add virtual servers as required without the need to purchase additional hardware. Upgrading hardware becomes a simple process. Removing the requirement of the Operating System needing to work directly with the hardware makes disaster recovery or replacement of failed servers simple.