Effects of virtualization and cloud computing on data center networks

6
Figure 3: When using a VDI configuration Virtual Connect technology, only a small amount of production
(protocol) and management traffic exit the Virtual Connect domain, thus optimizing the E/W traffic flow.
Cloud applications
Enterprise businesses are moving beyond server virtualization and VMs to embrace cloud-computing
environments. The solution stack for these enterprise cloud-computing environments often includes
infrastructure as a service (IaaS), platform as a service (PaaS), and software as a service (SaaS).
These “X as a service,” on-demand models require flexibility and the need for immediate growth. So,
enterprises often build private clouds on virtualization and large L2 domains to allow live migration of
VMs across as large a domain as possible. Cloud models, especially in service provider
environments, also require a multi-tenancy infrastructure to provide separate and secure services to
many customers simultaneously.
Some public cloud applications and service provider environments use cluster-like architectures, with
massively parallel workload and data distribution characteristics. Common cluster applications
include data aggregation and “big data” analytics applications like Hadoop, Needlebase, Platform
Computing’s Symphony software, or Vertica software. As a request comes in, a task scheduler
spawns multiple jobs to multiple servers—causing a flurry of network traffic that does not go up to the
network core but out to peer servers. Social networking sites use services like memcache for
distributing memory objects to alleviate database load and speed up performance. Applications like
Swift and services like Amazon’s Simple Storage Service (S3) distribute storage across multiple nodes.