To Grasp the Cloud—Fundamental Concepts - Deploying and Managing a Cloud Infrastructure - page 40

including all major OS offerings—Microsoft Windows, Mac OS X, and Linux variants. All
versions, including the free version, come with a primitive set of functionality that includes
a graphical management console and memory overcommitment functionality. Some of the
unique features of vSphere ESXi are described in the following sections.
Memory Overcommitment
Memory overcommitment works the same way our banking system works. The central bank
prints more money than it has physically available. Similarly, the ESXi engine would allocate
more memory to each of the virtual machine you fire up than what is physically available
on the server. This is a huge undertaking in terms of the risks involved. If the memory is not
properly managed, one or more virtual machines may crash. This is based on the fact that
most of the time, not all memory is being completely utilized by all the virtual machines
running on a server.
Also, peak usage patterns indicate that whenever one or more virtual machines are
completely utilizing their allocated memory, there would be virtual machines on the same
server that are lightly loaded. This fact, coupled with smart memory sharing between virtual
machines and data compression, enables the ESXi engine to allocate maybe up to 400 percent
more memory than what is physically available.
An Example of Memory Overcommitment
The ESXi engine may allocate 4 GB of memory to four virtual machines running on a
server with only 12 GB of physically installed memory. Whenever an instance needs the
additional “thinly created” memory, the ESXi engine shares available memory from other
virtual machines with the virtual machine that needs more, giving the impression of
boosted memory.
vSphere High Availability
HA is essential for any enterprise-grade application running in a virtualized server environ-
ment. It ensures that applications and their environment keep serving end users in case of
both physical (hardware) and OS (software) failure. vSphere HA ensures that whenever a
server goes down physically, all the virtual machines running on it are migrated to another
available server without interruption. If the OS fails on one or more virtual machines running
on a server, those virtual machines are restarted on the same server. This enabled offering
HA to end user applications without the need to maintain dedicated failover servers because
vSphere would utilize available resources on the current production servers for any failover
migration needs.