I still remember very clearly while attending a Windows 4.0
training in MCSE (Microsoft Certified Systems Engineer) curriculum,
some time in the year of 1999, in the chapter ‘How to
configure TCP/IP’, the instructor demonstrated with a
few mouse clicks to set up TCP/IP. He excitedly said, after a
few clicks, “this is what I call an OS (operating system),
click click click boom TCP/IP is all set, Unix is dead”. It
has been more than a decade since then, Unix still runs in
most high SLA (service Level Agreements) environments like
production, staging and DR (disaster recovery) .
While participating in cloud conversations, I get a similar feeling
when I hear most cloud computing vendors swearing on its success
and end of conventional hosting and private data centers.
Cloud computing for sure is very promising; however, for now on it
is just at the nascent stage. It has to go through all
maturity and stability curves, ecosystem readiness, before if can
be consumed the way it’s being sold today -- a default by
choice. It might not take that long to mature like technology did
in the 1990s or 2000 but I assume it will still take another 3-5
years.
Cloud computing is not just a piece of technology in the digital
world, but it will be a key building block for businesses to
function.
There are multiple parameters that can risk businesses consuming
cloud computing and make them run for their lives. Here I would
like to highlight what I think are the top three:
#1 Actual cost (Total cost of
ownership)
Comparison done by vendors for cloud computing to conventional
hosting (captive where hardware and software are bought and
managed) or managed hosting (leasing of h/w, software with or
without services), in most of the cases is not ‘apple’
to ‘apple’. The popular ‘pay as you go’
pricing model of cloud computing is only for the compute power and
that is CPU cycle (how long the CPU was engage in serving request),
Memory utilization, disk space consumption. Any thing beyond the
compute power like back up, data transfer, security, monitoring,
security, data life cycle etc are add-on services and have to be
paid separately. The sum of all the costs results into a
substantial number. If the application that will sit on the cloud
infrastructure have a 2-3+ year or have a road map for few years,
then the cash out flow in most probable cases will be higher than
the other hosting options The interesting aspect of these add on
services is that the services are provided by a third party and not
the cloud vendor. This introduces too many cooks in the kitchen and
managing multiple contracts and vendors is a nightmare.
#2 Cloud computing vendor
liability
Cloud infrastructure is setup by using hypervisor, which enables
the hardware usage to the maximum by sharing compute power (CPU,
memory) in the background. Technological advancement in hardware
has made servers smaller but more powerful. Hence lots can be done
by very small size of hardware (servers). Cloud vendors slice out
virtual machines leveraging hypervisor (like Red hat, HyperV or
VMware) to their customers. Customers will have full access and
control (install, reinstall, start, stop restarts, change
configuration) over their virtual machines but actual location of
the virtual machine where it resides, is rarely known. So in the
event of a hardware failure, the vendor support team would know
which hardware (physical server) has crashed but would have limited
knowledge or no knowledge of which customers’ virtual
machines are impacted. More importantly, to bring back the
application, the customer must either have an adept team in the
house to find a partner to administer technology above hypervisors
like OS, COTS (custom of the shelf) software.
The cloud is a multi-tenant environment where many clients share
one set of hardware infrastructure. Criticality of that hardware to
be available all the time is immense. In the event of failure of
the infrastructure, there will be an impact on multiple business
applications. As a result, depending on the type of business
application running on the shared infrastructure, the damage could
be of different categories (customer experience, money, loss of
business, brand etc) and scale.
In such a scenario, imagine the size of the liability the cloud
vendor will ride on when there are many clients on a shared
infrastructure. In case of a mishap the total sum of liabilities
could be more than the company (cloud vendor) is worth.
Cloud vendors hedge it by buying additional insurance. But
insurance folks, who are either reluctant or bump up the premium
that at the end of the day is passed on to the customer. That is
why the cloud vendors shy away from putting a clause for the loss
to business in their legal agreement and pay just the fee waiver of
the total down time, which in most cases would be inconsequential
as compared to the impact to business.
#3 Legality
Courts round the world are beginning to understand cyber law and
are busy writing law on cyber security. Cloud and its set up
complexity is a generation ahead of what court understands it
today. As cloud is virtual and span across boundaries, so when
there is a breach of contract or data is compromise or down time
incident then who and how the issue will be contested, which
land of law prevails and the issue is brought to justice.
Cloud in the most vanilla form, is nothing but virtualization. It
‘will’ become imperative to virtualize hence the need
to starting thinking and strategizing now.
For enterprise or SMB, CIO must have visibility of their
applications and infrastructure for the next few years, if not
please get a new CIO. They must think of creating a
virtualized infrastructure for all their applications. Start on a
smaller scale and gradually gain weight as and when your
application reaches the end of life or infrastructure needs to
rejuvenate. Doing it in the house will save tons of money, if
viewed on a long term basis (that is 3 years plus). This will keep
the risk of security, legality and liability at bay as all are
under your control and can be customized.
For smaller businesses that need to focus on laying down foundation
of their businesses, they should leverage cloud for instance
gratification and gains like – CAPEX, setup complexity, time
to market, flexibility to move around.
- Mohammad Wasim is Director at
Sapient