Virtualization is about to change the game again in the datacenter. As the modern computing world has become comfortable with cloud computing, their appetite is accelerating for it, and doing so rapidly. In fact, Gartner recently reported that spending on public cloud services will be $109 billion this year, up from $91 billion from last year’s spend. And it will nearly double to $207 billion by 2016. That’s a consistent increase of over 20% each year, and the fastest growing area of spend according to their predictions
And guess what? Some of them are in your business, and you probably don’t even know it. Analysts are calling this trend “shadow IT” where end users decide to implement their own CRM solution with a simple credit card swipe. Or where a business unit decides to build and test an app on Amazon instead of internally on your infrastructure. PricewaterhouseCoopers (PwC) has data that shows large enterprises (averaging $500MM in revenue a year) “leak” as much as 30% of their IT budget outside of IT’s purview and ledger. Not only do these costs surprise the CIO, but they also fall outside of IT’s ability to govern, secure and maintain compliance. And of course, when things go wrong, IT is drawn into help troubleshoot a solution they have little knowledge of, exhausting resources very quickly.
This is a problem IT and companies like VMware are working hard to turn the tide back towards where users want to look to IT first to solve their problems and see the value of keeping IT projects in-house.
The New Cloud Operating Model
In order to steward users through this climate change brought on by cloud computing, and in some cases, to even remain relevant, IT needs to transform itself into a service provider of cloud services that are more attractive than public cloud services.
But why fight it? The Aberdeen Group gave CIO’s a clear incentive last year, citing that the private cloud saves a total of 12% combined annual cost savings over public clouds on a per-application basis. Of course, this only improves with scale. Based on working with thousands of customers making the transition to virtualization, VMware recommends that any organization with over 50 virtual machines in the cloud should be hosting those virtual machines internally.
So what does IT need to do? IT needs to transform itself both organizationally and technically, and empower the business further to build, run and manage their applications on their own. IT will manage hardware, security, and corporate policies, while application teams will become DevOps teams, responsible for the application Service Level Agreements (SLAs).
This won’t come overnight, so let’s start thinking of it as a three phased approach.
Phase I: Infrastructure Focus
In a nutshell, this phase is dedicated to building out the foundation of your virtualization infrastructure. This isn’t as simple as setting up vSphere and calling it a day. Your IT organization needs to develop a competency in virtualization infrastructure administration (VI admin). This means that your VI admins need sophisticated management tools, like vCenter and vCenter Operations, so they can automate management and control of potentially thousands of machines. They to build investment and competence in virtualization monitoring tools, orchestration tools, and provisioning tools, like vCloud Director, that include the ability to “burst” to public clouds in times of high capacity.
Phase II: Application Focus
This phase is about building out the services that will attract your business units to load their apps on your infrastructure versus public cloud services. It starts with they want to be able to spin up a virtual machine whenever they want.
Just like public cloud’s web store, a self-service portal is the first start. Next, application owners need to be able to create designs for their applications (VMware calls these application blueprints) that they can save, update and redeploy over and over again—across development, test and production systems. Ideally, they can also choose to deploy these applications to public clouds in order to get projects started faster when resources are constrained or in case special projects demand too much for internal infrastructure (think the case of the New York Times Machine).
>> For more info on VMware’s product offering here see the VMware Service Manager and vFabric Application Director product page.
>> For more info on VMware’s product offering here see the vFabric Application Performance Manager product pageAt VMware, we’ve seen that to achieve a real DevOps separation from IT’s umbilical cord, you will also need to put in application centric monitoring and management tools that are built in to that self-service portal and available for DevOps to use independently. Since the DevOps teams will not have access to control the underlying infrastructure, it makes no sense for them to share IT’s tools. They need separate tools that are completely focused on how their applications are running across clouds. These app management tools need to be smart, they need to follow the application wherever it goes, and however it scales. They need to be change-aware, and need to look first look at transactional management to identify where applications are getting stuck. It should also monitor and manage middleware, providing app teams with insight to when configurations for the app server were changed, error logs for the web server, and the ability to run routine maintenance such as restarting the virtual machine itself directly from the management console. For those going for a gold medal in customer service, it will also include a byte code instrumentation capability that will help developers (the Dev in DevOps) to examine stack traces and fix code quickly.
This is the phase that most organizations need to get to right now to land apps on their infrastructure, and remain in control. If you need more convincing that these end user services will attract users and apps to your infrastructure, put yourself in a DevOps users seat and check out a 7 minute video by VMware that shows how this self-service provisioning and management process looks to them.
Phase III: Governance
This phase is likely a ways off for most organizations. But it is the envy of every CIO. It starts with building upon the service catalog that powered the self-service portal, and tracks costs. It tracks information from the software licensing costs, the labor costs to support the service, and the raw infrastructure costs such as hardware, power and cooling to run the service. It can then “charge back” usage to individual business units. This will help the entire organization to better rationalize what services to keep in-house vs outsourcing. By tracking external services in this model as well, it will also highlight new services that IT should be offering. This level of financial transparency will ultimately help business units to be more diligent about wasted services and not leave a test environment up for weeks that is only used for days. While tedious to track and manage, once running this phase will give the CIO and IT the information they need to rationalize cloud strategies that are right for them, and keep the business moving forward as fast as innovation takes them.
>> For more info on VMware’s product offering here see the VMware IT Business Management Suite™ product page
And for those IT veterans who think they would never have time to get down to that level of detail, just imagine how much time you will have once you are relieved of your DevOps responsibilities.