“Products and standards are not mature enough to provision
and manage large
clouds. While only Linux and Windows environments are supported by
most of the cloud service providers and cloud-related technology
providers, around 80
percent of business-critical enterprise applications are hosted on
Mainframe, Solaris, AIX, HP UX, and so on,” says R
Viswakumar, AVP, Global Technology Office, Cognizant. If, for any
reason, the customer decides to move from one cloud to another, the
applications built to run on one cloud may not run on the other and
would require the customer to go through the entire process of
rewriting the application again. In case an inter-application
communication (between applications that the customer has on
premises and those within the cloud; or applications hosted on
different clouds) is required, the communication may not always
happen.
Joshi of Patni believes that not all of an organization’s
computing tasks will move into the cloud at one time; some will
always remain in-house for various reasons. It is also likely that
those tasks that move into the cloud and those that remain behind
interact with each other very closely, exchange data or
transactions frequently, share common databases and depend on each
other’s successful execution since they represent discrete
steps in a single continuous business process.
Therefore, integrating the two—tasks that have moved into
the cloud and tasks that have remained behind—is a
non-trivial technical issue. Resolving this will require
significant effort that may require alterations or re-engineering
of applications and their interfaces. This may ultimately even
require a redesign of the underlying business process.
“It is absolutely necessary for a cloud vendor to provide
cloud services based on open standards and open source with no
hidden APIs. No cloud provider will be able to compete if he is not
open source and will miss the boat if he makes the customer rewrite
his applications to run on his cloud,” says Mike Evans,
Corporate Development, Red Hat.
Though there are companies working together to standardize
application, file, and data formats on the cloud, Evans believes
that these standard group efforts are by companies that are
proprietary in nature. Says Evans, “They will open up a few
APIs for customers to adopt their cloud, while keeping the rest
locked in proprietary standards and get customers to run their
applications and store data on these standards. Customers will have
trouble in accepting that five proprietary guys who otherwise hate
each other are going to work together towards open
standards.”
It becomes imperative for cloud vendors to have their cloud
services based on
open standards and formats that can be supported seamlessly across
different
platforms. If a customer has to go through the application
re-writing processes
each time he moves his data and applications on to the cloud, he
may not necessarily adopt the cloud. Those who already have their
data on the cloud may move away from the cloud if they are not
offered the desired level of flexibility. This would result in a
great loss to the vendor in terms of customer base.
Service Level Agreements
Service Level Agreement (SLA) in itself poses a challenge since
customers
already have apprehensions about the aforementioned risks when
migrating to a
cloud-based architecture. Having the infrastructure internal to the
organization
ensures that demands in terms of performance, security of
applications and data,
network speed and latency are met. However, if data and
applications are stored
in a third party data center, the vendor will be required to
provide an additional level of assurance that requirements will be
met, well within the costs. “Although it is not directly
relevant to the clients, most of them, particularly enterprise
customers, would like to understand the building blocks of the
offering such as the data center features, virtualization
technology used, storage used, backup and DR features and so on.
Lack of access to such information increases uncertainty for the
decision makers in the client organization,” says Ramkrishnan
of Wipro Infotech.
Evans of Red Hat echoes this point of view: “The
uncertainty over the availability of applications, data, their
management, performance measurement and how data and applications
will inter-operate, needs to be addressed for a cloud to succeed.
If not, then there will be bad case studies where the cloud shuts
down because these issues were not addressed and the customer was
not able to make a switch.”
Adds Ananth Krishnan of TCS, “Higher-order questions will
include service
levels from the providers, extent of service lock-in, data
confidentiality,
audit ability, performance, and latency for the end user. From an
application
perspective, the issues of truly exploiting the capabilities of
cloud infrastructures could need skills in large-scale distributed
and parallel computing architectures and concepts.”
Viswakumar of Cognizant says that other issues that could prevent
cloud adoption include the need for new and complex skill sets and
roles which need to be created and managed, changes required in
supplier and consumer ecosystems, high legacy migration costs, and
higher resistance from companies to moving out.
The Case For Public Versus Private
Cloud
In spite of the challenges surrounding the adoption of cloud
services, it is not really the end of the road for organizations
looking to adopt cloud services. To deploy services on a cloud, an
organization needs to first decide whether it really needs to
deploy cloud-based services. While the decision to deploy a
cloud-based service is generally driven by the need to reduce the
cost associated with growing storage or computing requirements, it
is also necessary to decide whether cloud services are the best
possible solution to an organization’s growing infrastructure
demands.
Santhosh D’Souza, CTO, Sun Microsystems India says that
large organizations such as banks often have several business units
running their business applications on respective IT infrastructure
in silos. This leads to an over-provisioning of resources which
cannot be efficiently used unless shared.
Also, there will be organizations that have legacy architectures
with well-defined workload parameters, no major fluctuation in
workloads, and which have an idea of scalability.
“Such organizations can standardize on a set of
technologies and deploy these resources within the data center
rationalized into a shared pool. While mission critical
applications can be set up on systems with faster storage and
computing, the rest of the resource repository can be sliced and
allocated to each business unit. By laying down a billing model,
the cost of using these resources can be billed and sent to
individual business units,” says D’Souza.
D’Souza also observes that while private clouds are best
suited to organizations with legacy infrastructure, public clouds
can be used by start-ups that are unsure of the success of the
business. The absence of any major investment in actual physical
hardware and the overall lower capital investment will make it easy
for these organizations to put forth a proposal for adopting cloud
services with venture capitalists investing in the
organization.
Public clouds can also be used by companies delivering products
and services over the Web, with no captured set of users using a
single network and software developers with dynamic resource
requirements. Services such as Storage-as-a-service can turn out to
be cheaper for social networking sites wherein the data is
generated and stored online, thereby not requiring these companies
to go through storage buying cycles.
Another aspect is the presence of a large mobile workforce in
organizations that need access to applications on-the-go. Depending
upon the criticality of the applications; the organization can have
these applications hosted on the public cloud and make them
available to its mobile workforce via the Web.
Taylor of Akamai explains this approach and says, "Organizations
will adopt hybrid clouds which are a combination of public and
private clouds. They will have public clouds for the nomadic users
within the organization who interact with data which is not so
sensitive in nature, and private clouds where they can predict
usage, have employees in a single location, have knowledge of their
branch offices and have certain security concerns about the
data.”
‘Cloud’ Is The New
Buzzword
Most analysts and vendors believe that India is still a nascent
market for
cloud-based services. However, with the large SME/SMB population
and the
emergence of many start-ups, the scope for growth is huge, provided
the concerns over the adoption of these services are addressed.
While there is a certain level of awareness about what services and
benefits the cloud has to offer, organizations are taking a
cautious approach and are still experimenting and testing both
public and private cloud services. CIOs too share a similar point
of view.