MicroStrategy is a leading provider of Business Intelligence
(BI) and enterprise software with a direct presence in over twenty countries. Their
software speaks 14 different languages and runs on a variety of operating
systems including Windows and Linux.
The company went to a private cloud infrastructure to
expedite machine builds for over 1,000 engineers working in an agile
environment.
Business challenges meet the private cloud
Alex Freixas, vice president of worldwide information
systems at MicroStrategy, says that the company faced a need to expedite operations
and increase productivity across the company’s technology and R & D
departments.
“Every night we compile our source code in a single central
repository,” says Freixas. “We integrate our code every day. Initially, our
method was wipe out and rebuild machines for staff to get the latest build.”
MicroStrategy’s manner of code integration raised the
question of whether they could accelerate the pace of provisioning environments
for their engineers. Could they do away with software engineers and developers
having more than one workstation or using decentralized methods of virtualizing
development environments?
Freixas says: “We wanted to make sure that our entire staff
involved in the development were fully focused on doing what they were good at.
That is, developing code, testing code, and validating code and not necessarily
spending cycles provisioning machines.” They were out to remove the necessary,
but overhead, task of rebuilding machines from their 1,000 engineers’ plates,
allowing them to focus on developing leading edge BI solutions.
MicroStrategy wanted their technical support staff focused
on customers, not preparing environments. “It used to be tedious and time
consuming for our technical support engineers to replicate customer
environments so they could help troubleshoot issues in regards to our software,”
according to Freixas.
These business challenges led them to start looking at
building a private cloud to help accelerate machine builds.
Technology selection process
“A couple of years ago, we were relying quite a bit on
public clouds for provisioning of
environments for our engineers,”
according to Freixas. “Essentially as time went on, we realized the cost was
quite high in doing this on a massive scale with the public cloud.”
“We decided to look at what we could do to continue
improving upon our operational efficiencies but at a lower cost, given that we
have the requirement at any time to provision a few thousand machines,”
according to Freixas.
MicroStrategy’s history of working with private cloud type
software with VMware goes back to 2008. At some point, they started hitting
scalability issues.
“That’s why we were using a public cloud provider as well–because
we couldn’t satisfy the computing demands internally,” according to Freixas.
The folks at MicroStrategy spent a good part of 2009 and
2010 evaluating their option–did they want to continue relying on public
resources to satisfy their computing requirements, or should they design a new piece
of infrastructure from zero? For the latter, they spent about six months
evaluating best of breed and what would work in their case in terms of
computing, storage, network, and hypervisor.
They wanted to have a highly automated, open cloud against
which they could develop code. Self-service, as well as integration with Active
Directory, was also key to them.
“The VMware stack is unique in terms of management,
automation. It’s not only about the hypervisor but also about the
administration and self-service capabilities you get running the enterprise
version of the vCloud suite.”
Management, monitoring, and automation factored into their
technology selection process.
Efficiency gains
Server consolidation is an obvious boost. Freixas shares, “We
were able to displace 44 racks worth of data center space (equipment and
servers) onto the cloud. We were able to virtualize 95 percent of the R&D
Server infrastructure.” Those 44 racks were reduced to exactly 21 square feet
of data center space.
Reducing MicroStrategy’s datacenter racks in this way meant
tremendous cost savings in terms of cooling power and server administration.
“Not using external cloud providers provides significant
savings,” says Freixas. “At this point, I would say we’re saving over a million
dollars a year from cost we’re not incurring anymore.”
Automated provisioning machines also represent some
incredible time savings and productivity gains for their software engineers,
developers, and technical support staff. Their private cloud infrastructure
saves 45 minutes per engineer per day, and in the past twelve months they’ve
been able to deploy 14,000 machines through automation from their private cloud.
“We started off with our private cloud by provisioning
machines for our R&D department and all of a sudden it became vital,”
according to Freixas “Today we are serving our technical support organization
in the same cloud.” Now the company is also provisioning training classrooms in
their customer training facility.
MicroStrategy also uses their private cloud to accelerate sales
activities by deploying customer proofs of concept in a matter of minutes
without having to rely on customers or prospects to provision an environment.
Lessons learned
“You have to make sure you do things right,” advises
Freixas. “Spending ample time evaluating alternative technologies and building
a reference architecture was key to success for us,” advises Freixas.
He further advises, “You have to design it right. You have
to design your cloud with expansion and resilience and high performance in mind.
Building a cloud isn’t a matter of getting the VMware software and populating
it in a series of servers. Design is
definitely key.”
He also cites the importance of what he calls “the third
generation of IT engineers.” Freixas says, “Your network engineers have to
think virtual, so there has been a shift of mentality in our worldwide IT
function where engineers actually take virtualization
very seriously.”
He also recommends working with end users when making such a
private cloud move.
Conclusion
Freixas says, “After we were able to successfully deploy our
private cloud, our engineers got their life back.”
MicroStrategy’s using their private cloud to offload the
time-consuming task of rebuilding machines helps the company maintain a
productive and agile development environment that certainly contributes to the
company’s bottom line.