Understanding Cloud Characteristics - Deploying and Managing a Cloud Infrastructure - page 11

Metering Pooled Resources
Consider this scenario to understand the complexity in metering pooled resources. When a
physical machine is virtualized, it enables spinning up and shutting down multiple operating
system (OS) instances almost in real time. Although virtualization providers like VMware do
provide mechanisms to allocate resources on a server (compute, storage, and network) at a
granular level, there would be overflow based on the number of VMs running on a physical
server at any given moment. If a dual Intel Xeon server with IB ports and SSD storage has
just a single VM running, the application running on the VM would definitely give its best
performance. However, when the server gets saturated with the maximum number of VMs
that may be spun up, the resources would be strictly rationed.
The cloud vendor would need to have metering built in so it can account for pooled resource
utilization at the atomic level. This resource metering would have to implemented on top of
the VM resource monitoring and not at the level of bare metal.
Multitenancy
Data centers of cloud vendors are deeply virtualized, which translates into the ability of
the vendor to pack multiple users or tenants into the same physical servers. This is what's
referred to as multitenancy within the cloud.
Single-Instance Model
Multitenancy is often used with the term single-instance model . They both refer to the
same feature of the popular definition of the cloud where physical resources are virtualized;
the physical layer is completely abstracted and offered as billable units of compute, storage,
and network resources. In this model, the applications of multiple tenants belonging to dif-
ferent companies reside on the same physical server but are segregated at the VM level so
that data, software, and custom applications and configurations running on one tenant's
cloud account are not accessible to the other tenant. This segregation is implemented within
the virtualization layer, which ensures that data and access cannot leak between multiple
VMs running on the same server.
Customized Configurations
Every application has unique requirements for tweaking the software it's running on. This
includes, for example, configuring a web server running on top of the cloud VM allocated for
the application. Multitenancy would not stop a tenant from customizing software running