Cloud Management - Deploying and Managing a Cloud Infrastructure - page 99

When preparing the network and IP documentation, you should consider all communi-
cations occurring in and out of the network: between the data center or local network and
its users, within the data center itself or within the network, and between the data center or
network and other networks, clouds, or data centers. Detailed documentation of each type
of communication and how they connect with each other, especially in terms of hardware
mapping, will aid in root-cause analysis of network problems and help in making informed
decisions in future network upgrades. The network is a combination of three domains,
which are integrated differently depending on the customer's needs:
Core Network The core network provides a high-speed packet-switching fabric for multiple
access and edge network devices as well as high-speed communication within the local system.
Access Network This is what most internal users will be on. It provides connectivity to
shared enterprise servers and network-attached storage devices and any other IP or auto-
mation devices required by the system.
Edge Network This provides connectivity to external users from different network types,
most commonly the Internet. Wide area networks (WANs), backbones, and virtual private
networks (VPNs) are other networks that can connect to the edge network.
Application-Optimized Traffic Flows
In the old client-server architecture of networking, traffic was in effect a single north-south
direction scheme. Requests originated from a client system and went directly to an application
server that responded directly to the client. But with the increasing use of cloud computing,
interserver communication is now very common. In today's system, the client request is sent to
an application, but the processing of the request results in information sharing between servers
that exist on different machines and are distributed in different locations. And depending on
the request, multiple servers might be processing the request at the same time.
Documenting how data flows through the network would help in root-cause analysis of
network-related issues or conflicts. For example, a request might be approved, but it never
gets carried out because somewhere along the line a server might have dropped it for some
unknown reason. With proper data flow documentation, network engineers can follow the
data to see where the request gets dropped and fix the problem quickly. Events such as these,
that occur only intermittently and in very specific circumstances, are often the hardest to
track and often left unfixed because there are only a few users. Sometimes the IT staff tells
those users to simply change their settings, for example, instead of addressing the root cause
of the issue.
Simplified Network Infrastructure
Cloud computing requires homogeneous hardware and infrastructure to provide scale and
elasticity for rapid growth while also keeping down costs. Old network architecture that
previously worked well for fragmented and oversubscribed tree structures and business pro-
cess will have consistent scaling and performance issues. A simplified network infrastructure
meets all the needs for scaling, performance, and cost effectiveness.