When we’re on the customer
trail discussing vCloud solutions and architectures, the cloud computing
phenomena often presents itself as a series of executive briefing anecdotes.
One of our favorites is a senior IT leader at a major bank who through an
audit discovered more than 5,000 virtual machines being consumed through public
cloud services – despite hosting more than 50,000 servers behind the corporate
How can an organization
with such large-scale capacity not be fulfilling the needs of its internal
customers and developers? Why is cloud computing now accounting for 10 percent
of the company's server consumption? Simple differences in consumption and costing
models, spread across a large organization, can add up.
Software development and
testing is a great example. Servers and infrastructure resources are often
allocated like raises – once you get them they are rarely taken away – and so
like raises they tend to have an approval process with a tension between need
however, can be iterative, experimental and demanding. Often developing a
complex Java application can require far more resources than operating it. The
development environment requires a full instantiation of the application, but
adds additional resources to put the application under a demand load. Cloud-based
test leader Soasta has already begun to capitalize on this need – but not every
application is suitable to its web application focused solution.
But taking a step back the
familiar refrain of "It spends 75% of its budget just keeping the past
running" is important. If the way to get ahead in corporate IT is keeping
the lights on (availability) that means the application development can fall
slightly lower on the resource provisioning priority list – despite having a
greater demand on server resources compared to full running production applications.
The cloud consumption model – pay as you go, immediate provisioning, able to
create a large performance-testing farm – is powerfully aligned with the
workflow of application development.
But cloud is going even further than being just a consumption model. As server request and configuration makes its way even closer to the developer's toolkit with projects such as jCloud and libloud, developers can now automate even more of their testing environments. This combination of new methods of code level automation, mixed with a flexible consumption model, is part of what makes the cloud the most compelling early home for new software projects.
If you want to get more information on the progress
of the vCloud API, check out http://communities.vmware.com/community/developer/forums/vcloudapi,
follow us on Twitter at @vCloud or become a fan of our Facebook Page at http://www.facebook.com/#/pages/VMware-vCloud/203532122896?ref=ts