Cloud Source
Showing results for 
Search instead for 
Do you mean 

Implement Cloud, one step at the time

on ‎08-08-2014 08:52 AM

cloud maturity.pngAs Cloud has become mainstream, and as business organizations have started using it, many IT departments seem to rush to the Cloud. Does this makes sense as Cloud is only one step in an evolution of the IT landscape to a cheaper and more agile environment? Let’s put this evolution in perspective and let’s describe 5 key steps along the way.


Step 1: Standardize and Consolidate

Many companies have started virtualizing their environments. The objective being to run more applications (in the Cloud jargon they are called workloads) on the same server without interactions between them.


Twenty years ago, systems were multi-tasking – running several applications within the same operating system environment. But as applications became more complex, there was a need for each application to have its own operating system and middleware environment. So there became a need to encapsulate them within a Virtual Machine (VM), running on ever-more powerful hardware.


And VMs have been created by the thousands to the great advantage of the hypervisor companies, but not always to the one of the IT budgets. We have known of “VM sprawl.”


Each CPU has a number of Virtual Machines that can run on it. They are stored on their hard disks, and uploaded when required. It improves the use of the servers, but does not improve flexibility. Standardizing and consolidating infrastructure (Servers & storage) should actually have been the first step. By standardizing on one or a couple platforms, Virtual Machines can now easily be moved from one physical server to another when required. Unfortunately, most companies have not done the standardization and consolidation.


Step 2: Virtualization and Automation

As mentioned, virtualization is often done at this point. Actually, I have to be more precise. Server virtualization is done, storage and network virtualization are still required to create the more flexible environment we talked about. The objective is to optimize the use of the hardware and improve the management of workloads.


In the meantime, unrelated to this, the number of operating systems, middleware and – in particular – application updates has increased. Indeed, to match the business needs more closely and to improve agility, developers tend to move away from the traditional release cycles to more agile development methodologies.

This results in additional work for the operations team, as they need to generate new Virtual Machine images much more regularly than before. And here is where automation adds value. Why not automate all the routine tasks, albeit to install the applications, set-up or maintain the environments? In many enterprises I visit, this last step has not been fully embraced yet. It is, however, a pre-requisite to implement Cloud-type environments.


Steps 1 and 2 are fundamentally infrastructure focused. Their objectives is to get the infrastructure more productive without exploding the cost of managing it. Things are still under the firm control of the IT department in general, and the operations people in particular. They often become the bottleneck in the quest for more agility and responsiveness. This is why the question often arises as to whether the requestor could initiate the process him- or herself. But that is a huge change leading us to step 3.


Step 3: Self-Service Infrastructure

Could Could application developers, or the team managing applications, automatically request the appropriate infrastructure they require to run their environments? It would speed-up the development and application roll-out process, while reducing the amount of mundane tasks the operations team has to perform. Obviously, that team wants to make sure things are done according to the book – hence, the importance of automation. You don’t want the requestors to have to manually set-up the environments, to run the scripts, to configure the devices, and etc. To make sure it’s all done correctly, you want this done automatically. This is what is called (in Cloud jargon), Infrastructure as a Service (IaaS).


As developers often require multiple environments along the development/testing process, giving them the opportunity to self-provision those environments provides two key benefits. First, they no longer have to manage their own development machines, and at HP, we realized they typically spent 15 percent of their time doing that. So, self-provisioning makes them more productive. It also allows the operations team to define which version(s) of the operating system, databases and middleware are acceptable. These are the only versions for which virtual machine images are provided to the developers. This often results in a drastic limitation of the amount of versions in use, facilitating maintenance and reducing support costs. It also helps managing license costs.


I’ve often seen companies making their first steps in Cloud through the provisioning of infrastructure by the development teams. This allows companies to experience the benefits of Cloud in a controlled environment and learn along the way the implications of introducing Cloud Computing in their infrastructure portfolio. The difficulty (for the operations people) is to accept they are no longer in full control of every aspect of the use of the infrastructure they manage. They now provide a service to other teams, and their responsibility is to ensure the service is always up and running at a satisfactory level.


Step 4: Self-Service Applications with Full Lifecycle Management

Larger companies typically have thousands of applications. Not all of them need to run 24/7. Some are only occasionally used. So, why have those applications consuming resources and license keys when they are not in use? Could we not allow the end-users to provision them when required and de-provision them? In other words, release the resources consumed, when no longer needed? This is application, or as it is called in Cloud terms, Service Provisioning. Business users, potentially with the approval of their managers or according to some other rules, now request access to applications themselves, freeing the help desk from this task. They run the application as long as they need it, and release it.


For IT this may be nerve-wracking as they now deliver a service to the business teams. In many cases they do not have a good understanding of the actual demand, so – particularly in the early phases – they will have to be prepared to intervene on short notice to ensure the business users receive appropriate service. Again, automation is key here, as is integrated (sometimes called vertical), operations management. It is no longer enough to look at how the infrastructure works. It’s now important to understand the complete stack, from infrastructure to application. Business and IT processes should be standardized and automated. Service Level Agreements should be defined and closely monitored.


Step 5: Become a Service Broker in a Hybrid Environment

Up till now, I have mainly talked about the use of one Cloud environment, typically a private or a virtual private Cloud. But we all know that one Cloud typically does not address all user needs. I started this discussion pointing out business organizations have started using Cloud environments (SaaS and Public Cloud typically). Once they have gotten used to provisioning their applications in the enterprise Cloud, they are doomed to come back and request the integration of those SaaS and Public Cloud services with the Cloud environment proposed by IT. Being able to provision and access those external services from the same portal, and being able to integrate them with enterprise data residing in the IT Cloud environment, are two of the key questions that will come up. This is where the concept of service broker plays. Your enterprise Cloud becomes the single place to which the business users go, and through which they request and access all their services/application/data needs. All services are provided through a customized and unified environment.


The benefit for IT is that they regain control over all services used within the enterprise. The disadvantage is that they are now expected to manage those suppliers. Often the costs of those services suddenly also hits their budgets. And that is where things become more difficult. Indeed, most IT departments work from fixed budgets and are supposed to deliver all IT services from that budget. But consumption of external Cloud services is a variable component, not fully controlled by IT as the requests directly come from the business.

These aspects have to be discussed and addressed upfront to ensure appropriate approaches are taken and the business users feel responsible for the consumption of IT services.


Taking it One Step at a Time

Where step 1 and 2 were mainly infrastructure focused, the last three steps are clearly services focused. Obviously there are many questions around the integration, management and end-to-end security of these environments that need to be reviewed. But that will be for another blog entry.


The business is looking for increased agility and responsiveness. The financial teams want to reduce the cost of IT. Standardizing, virtualizing, automating and providing self-provisioning are the ways to go and address the needs of the business and the CFO. Take it one step at a time though, as the technology exists. But the organizational, financial and human aspects should not be forgotten. To be successful, do not forget management of change, as an IT department that has gotten through these five steps looks very different from when it started.

About the Author


Nov 29 - Dec 1
Discover 2016 London
Learn how to thrive in a world of digital transformation at our biggest event of the year, Discover 2016 London, November 29 - December 1.
Read more
Each Month in 2016
Software Expert Days - 2016
Join us online to talk directly with our Software experts during online Expert Days. Find information here about past, current, and upcoming Expert Da...
Read more
View all