114
INTELLIGENT CITIES
series of noteworthy technology-inspired transformations, optimiza-
tions, and so on. e next in line is the availability and adoption of
several tools to automate many manual activities. Automated capacity
planning, job scheduling, billing and charging, resource provisioning
and deprovisioning, resource monitoring, performance measurement,
governance, self-service, and so on are being activated through highly
efficient configuration and management tools. ese enhancements
have laid the foundation for the flourishing of cloud centers in prime
locations across the world.
Having tasted the expected successes on server machines, the focus
has been directed toward storage and network virtualizations. e
tried and tested abstraction and decoupling have been the hallmarks
for virtualization. Further, all the hardware-bound intelligence is
smoothly extricated and developed in centralized and clustered con-
trollers that are extensively filled with software for attaining exten-
sibility, modifiability, and sustainability. Controllers facilitate policy
and configuration changes, remote access, policy-based operations,
and so on. e Gartner Group [12] breaks down the software-defined
data center into four processes.
• Abstraction—the decoupling of a resource from the consumer
of the resource
• Instrumentation—the process of opening up the decoupled
infrastructure elements with programmatic interfaces (typi-
cally XML-based RESTful APIs)
• Automation—the use of APIs to wire the exposed elements
using scripts and other automation tools to remove “human
middleware”
• Orchestration—the automation of provisioning through link-
ages to policy-driven orchestration systems
ese four processes are the keys to the data center of the future—
a data center defined and controlled by software. is new paradigm
removes existing barriers related to the management of server, stor-
age, and network resources. Changes to infrastructure that used to
require days and weeks to implement can now be made in minutes.
In this new era for IT, application workloads are not tied to dedicated
servers, storage, and network resources. Instead, they move dynami-
cally to balance workloads and optimize infrastructure utilization.
115
THE ROLE OF CLOUD INFRASTRUCTURES
3.6 Building Blocks of Software-Defined Data Centers
e software-defined data center (SDDC) encompasses software-
defined compute, storage, and networking components. e sub-
stantially matured server virtualization leads to the realization of
software-defined compute machines. Highly intelligent hypervisors
(alternatively recognized as virtual machine monitors [VMMs]) act
as the perfect software solution to take care of the creation, provision-
ing, deprovisioning, live-in migration, decommissioning of compute
machines (virtual machines, baremetal servers, etc.), and so on. In the
sections that follow, we focus on software-defined networking and
storage.
3.6.1 Software-Defined Networking
e emerging technology trends indicate that networks and net-
work management are bound to change forever. Today’s data centers
extensively use physical switches and appliances that haven’t yet been
virtualized and are statically provisioned. Further, the current envi-
ronment mandates significant and certified expertise in each vendor’s
equipment and lack of an API ecosystem toward the envisioned pro-
grammable networks. It is quite difficult to bring in the expected auto-
mations (resource provisioning, scaling, etc.) on the currently running
inflexible, monolithic, and closed network modules. e result is the
underutilization of expensive network equipment. In addition, the
cost for employing highly educated and experienced network admin-
istrators is on the higher side. us besides bringing in a bevy of prag-
matic yet frugal innovations in the networking arena, the mandate for
substantially reducing the capital as well as the operational expenses
incurred by the traditional network architecture is playing in the
minds of technical professionals and business executives.
As virtualization has been contributing immensely to server consol-
idation and optimization, the idea of network virtualization has picked
up in the recent past. e virtualization aspect on the networking side
takes a different route compared to the matured server virtualization.
e extraction and centralization of network intelligence embed-
ded inside all kinds of network appliances such as routers, switches,
and so on brings a number of strategic advantages toward consum-
able and cognitive networks. e policy-setting, configuration, and
116
INTELLIGENT CITIES
maneuvering activities are being activated through software librar-
ies that are modular, service-oriented, and centralized in a controller
module and hence the new terminology “software-defined network-
ing” (SDN) has become hugely popular. at is, instead of managing
network assets separately using separate interfaces, they are controlled
collectively through a comprehensive, easy-to-use, and fine-grained
interface. e API approach has the intrinsic capability of putting a
stimulating and sustainable foundation for all kinds of IT resources
and assets to be easily discoverable, accessible, usable, and compos-
able. Hardware infrastructure programming and thereby their remote
manipulations and machinations are gaining momentum.
erefore standards-compliant SDN controllers provide a widely
adopted API ecosystem that can be used to centrally control multiple
devices in different layers. Such an abstracted and centralized approach
offers many strategically significant improvements over traditional net-
working approaches. For instance, it becomes possible to completely
decouple the network’s control plane and its data plane. e control plane
runs in a cluster setup and can configure all kinds of data plane switches
and routers to support business goals as demanded. at means data flow
is regulated at the network level in an efficient manner. Data can be sent
where it is needed or blocked if it is deemed a security threat.
A detached and deft software implementation of the configuration
and controlling aspects of network elements also means that existing
policies can be refurbished whereas newer policies can be created and
inserted on demand to enable all the associated network devices to
behave in a situation-aware manner. As we all know, policy estab-
lishment and enforcement are the proven mechanism to bring in the
required versatility in network operations. If a particular applica-
tion’s flow unexpectedly needs more bandwidth, the SDN control-
ler proactively recognizes the brewing requirement in real time and
accordingly reroutes the data flow. Precisely speaking, the physical
constraints are being eliminated through software-defined network-
ing. If a security appliance needs to be inserted between two tiers,
it is easily accomplished without altering anything at the infrastruc-
ture level. Another interesting factor is the most recent phenomenon
of “bring your own device” (BYOD). All kinds of employees’ own
devices can be automatically configured, accordingly authorized, and
made ready to access the enterprise’s network anywhere, anytime.
117
THE ROLE OF CLOUD INFRASTRUCTURES
3.6.2 e Key Motivations for SDN
In the IT world, there are several trends mandating the immediate
recognition and judicious adoption of SDN. Cloud-enabled data cen-
ters (CeDCs) are being established in different cool locations across
the globe to provide scores of orchestrated cloud services to worldwide
businesses and individuals over the Internet on a subscription basis.
Application and database servers besides integration middleware
solutions are increasingly distributed whereas the governance and the
management of distributed resources are being accomplished in a cen-
tralized manner to avail the much-needed single point of view (SPoV).
Because of the hugeness of data centers, data traffic, both internally
as well as externally, is exploding these days. Flexible traffic manage-
ment and bandwidth on demand are the emerging requirements.
e consumerization of IT is another gripping trend. Enterprise
users and executives are being increasingly assisted by a bevy of gad-
gets and gizmos such as smart phones, laptops, tablets, wearables,
and so on in their daily chores. As noted elsewhere, the BYOD
movement requires enterprise networks to inherently support policy-
based adjustment, amenability, and amelioration to support users’
devices dynamically. BDA has a telling effect on IT networks, espe-
cially on data storage and transmission. e proprietary nature of
network solutions from worldwide product vendors also plays an
unhealthy role in traditional networks and hence there is a clarion
call for bringing in necessary advancements in network architec-
ture. Programmable networks are therefore the viable and venerable
answer to bring the desired flexibility and optimization to corporate
networks. e structural limitations of conventional networks are
being overcome with network programming. e growing complex-
ity of traditional networks leads to stasis. at is, adding or releasing
devices and incorporating network-related policies are truly difficult
in the current setup.
As per the leading market watchers, researchers, and analysts, SDN
marks the largest business opportunity in the networking industry
since its inception. Recent reports estimate the business impact tied to
SDN could be as high as $35 billion by 2018, which represents nearly
40% of the overall networking industry. e future of networking
will rely more and more on software, which will accelerate the pace of
118
INTELLIGENT CITIES
innovation for networks as it has in the computing and storage domains
(explained later). SDN has the capacity to transform today’s static and
unhealthy networks into calculative and cognitive platforms with the
intrinsic intelligence to anticipate and allocate resources dynamically;
the scale to support enormous data centers; and the virtualization
needed to support workload-optimized, converged, orchestrated, and
highly automated cloud environments. With its many advantages and
astonishing industry momentum, SDN is on the way to become the
new norm for cloud and corporate networks.
3.6.3 Tending toward Network Virtualization
With the cloud era developing quickly, a greater concern is being
expressed about the slow pace of innovations on the networking
front. ough virtual machines can be provided on demand and the
resources such as processing power and storage can be added to them
dynamically as needed, there is often no optimized control of net-
work bandwidth. As a result, services can easily become starved for
bandwidth, resulting in a sharp decline in their performance. ere
are automations here and there but the end-to-end automation is the
need of the hour toward virtual computing.
Because of server virtualization, there are hundreds of computing
nodes that need to be cleverly networked and managed. Networks
today are statically provisioned with devices that are managed at a
box-level scale and mostly underutilized. SDN enables end-to-end
network equipment provisioning, reducing the network provisioning
time from days to minutes and distributing data flows more evenly
across the fabric, resulting in better utilization. SDN offers the pros-
pect of making network resources as fluid and network management
as centralized and automated as the rest of the cloud. Organizations
benefit immensely as SDN enables network administrators to con-
trol network traffic centrally through programming instead of relying
on more manual approaches; eliminates vendor lock-in for network
products because of its open and vendor-neutral software; makes it
possible to provide new services and applications quickly; reduces
operational costs because of its simple and automated approach to
deployment; and automates multitier system configuration and opti-
mization including the network appliances used between the tiers.
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.