Employees with something to lose will try to thwart your private cloud plans. The boss needs to take charge.
In most IT organizations, where infrastructure folks have something to lose in the move to the cloud, you're going to have to find the true believers, people who are excited about the benefits and don't mind doing some work to get there. Give them room to explore, which means taking other tasks off their plates and/or providing resources.
Buy some expertise. Even true believers don't have all of the answers. Joe Emison, an expert I spoke with for our cloud ROI report, maintains that "anyone who has ever successfully administered a Linux box can stand up a private cloud." But I've seen some very smart people have at least some difficulty getting going.
Maybe that's because there are so many cloud options. Open source? Proprietary? Do you use CloudStack? OpenStack? Eucalyptus? Nimbula? Geez, do you even use a software solution, or do you seek comfort in the expensive but accountable arms of a hardware/software Vblock solution?
You can't possibly go cloud at enterprise scale without a management platform, so which one do you use? Rightscale? Cloudscaling? enStratus? See what I mean?
All of a sudden there are lots of permutations. Spending a few thousand bucks on some outside expertise turns into a bargain compared to hundreds of hours of internal staff R&D.
One thing is clear. The truly successful private cloud implementations are typically using commodity hardware, not highly expensive hardware or hypervisors. As Nishan Sathyanarayan, director of professional services for Nimbula, points out, if you're satisfied with Google, Amazon, and Facebook, then you're satisfied with large-scale infrastructure running open source on commodity equipment.
I am well aware that Cisco and friends have invested heavily in VCE and the Vblock platform. They're trading on their brand names, on the fact that they have earned deep loyalties in the infrastructure world. But cloud is not about name brand infrastructure. Don't listen to your staffers when they tell you that the expertise they need is centered on brand name hardware and hypervisor solutions. It may be one day, but it's not now.
Want the secret sauce recipe? Find folks who know the software side, and who know how to use commodity gear. Get your training and consulting from them, not from the name-brand hardware people who have a vested interest in recapturing your enterprise's infrastructure business.
>> Start small, avoid academic exercises. If you have some cloud-obsessed folks on staff who have spent the time to come up to speed on their own, fine. Implement a small private cloud yourself. But be prepared for staffers to spend more time researching and debugging.
The bottom line is that your IT organization needs to experience a private cloud platform in order to make decisions about operations and maintenance and plan for what the infrastructure--and its staffing--will look like in 24 or 36 months. Even if you need to bring in outside expertise, you can get started for less than $20K.
But make sure that it's not just an academic exercise. Even if your initial private cloud is limited in scope, put it into production. You're going to be living with something like it in the future, so you'd better get used to it now.
Jonathan Feldman is a contributing editor for InformationWeek and director of IT services for a rapidly growing city in North Carolina. Write to him at firstname.lastname@example.org or at @_jfeldman.
InformationWeek is conducting a survey on IT spending priorities. Upon completion of our survey, you will be eligible to enter a drawing to receive an 16-GB Apple iPad 2. Take our IT Spending Survey now. Survey ends March 9.
Multicloud Infrastructure & Application ManagementEnterprise cloud adoption has evolved to the point where hybrid public/private cloud designs and use of multiple providers is common. Who among us has mastered provisioning resources in different clouds; allocating the right resources to each application; assigning applications to the "best" cloud provider based on performance or reliability requirements.