IT Service Management Blog
Follow information regarding IT Service Management via this blog.

Every cloud has an infra lining - on managing heterogeneous data centers - Part 2

This part will deal with private clouds and what hides underneath them. While the cloud based service delivery approach removes a limitation on the speed and efficiency by which services can go from idea to deployment it puts additional pressure on the underlying infrastructure. If managing change and configuration in enterprise data centers was tough before the cloud era now it becomes downright close to impossible. We have to remember that the challenges that existed before like the volume and frequency of changes is only exacerbated because now anyone can allocate IT resources. Trying to pull everything into the traditional change process won’t do since it will slow down the process significantly, thereby, taking away one of the main motivations for cloud – delivery speed. Changes that take seconds to implement will take weeks to review and approve. That simply won’t do.


Except for when consuming public clouds where the underlying hardware and software are managed by the service provider, IT organizations have to remember this point.

Let’s examine some of the challenges that a private cloud poses to the operations team:


Applying an OS security patch to physical machines The application of security patches is done on all machines with a certain OS regardless of their role. It’s important to understand the role of a physical machine in order to schedule and prioritize the action. One such role will be using a machine as cloud infrastructure running its virtual servers. The impact analysis run on the servers that will be changed must be aware of their role and the services they are working for. Since the data center will consist of legacy hardware and cloud supporting hardware, a system that does this impact analysis should be able to support both.


Data center maintenance – turning 30% of data center capacity off - Let’s say that maintenance of a data center’s air conditioning system has to take place and that as a result, 30% of its capacity has to be turned off for a period of time. It might even be migrated to another data center. A decision has to be made about which infrastructure can be turned off based on the services it is supporting and the timing during which this would incur minimal risk. The data center consists of both legacy systems and a private cloud. The decision making process about what parts of the infrastructure should be turned off has to have a full view of the entire infrastructure. It has to consider the services that the infrastructure supports therefore, a mapping of the infrastructure to its supported services. The decision can’t be made by looking at partial views of the environment. What if there are several private clouds each serving a different set of customers (both internal and external)? The view of the data center has to remain uniform.


Handling massive volumes of data center changes - Running a solid change and configuration management process on a highly volatile environment with high levels of employment of automation and virtualization such as we expect the cloud supporting one to be, requires a shift in thinking. The cloud disruption will cause more pressure to be put on the operations team managing the infrastructure that supports the private cloud. If times were hard before with the change team not really covering everything that goes on in production, think about a highly volatile environment as the one that the cloud delivery method will introduce. Automation cannot exist solely in the operations we do to the data center. It has to exist in the processes that run IT as well.


Virtualization bottlenecks - In an environment that supports the cloud there will be segments that are each autonomous in their freedom for internal changes within certain boundaries. Simply put, I can allocate, for example, X amount of computing power via a IaaS model and allow services supported by this X to change the allocation between themselves while not exceeding X. Each such segment is supported by physical infrastructure. However, bottlenecks will always exist somewhere unless we can divide the allocation across joint silos of servers, network and storage. Naturally, this approach has its own limitations and barring a technological breakthrough, such bottlenecks have to be monitored and handled.


Management at the service level – Now that some of your services are deployed to clouds and others are on your legacy infrastructure, do you want to manage your services using tools from different vendors? For example, when examining the availability of your services, their TCO etc’ you will naturally want to be able to compare apples to apples using the same tools.


Controlling your policies – As data centers become more standard and as the automation becomes more pervasive, the act of changing a policy becomes a potentially disruptive activity in itself. Because such an action could have a significant impact on your data center as the updated policy generates new breaches that are handled quickly by automation tools, a process for controlling such changes is required.


Enterprise IT customers should be aware of these coming challenges and look for products that will allow to handle them.

Leave a Comment

We encourage you to share your comments on this post. Comments are moderated and will be reviewed
and posted as promptly as possible during regular business hours

To ensure your comment is published, be sure to follow the Community Guidelines.

Be sure to enter a unique name. You can't reuse a name that's already in use.
Be sure to enter a unique email address. You can't reuse an email address that's already in use.
Type the characters you see in the picture above.Type the words you hear.
Showing results for 
Search instead for 
Do you mean 
About the Author
I'm a 20 year veteran in the software industry with experience ranging from control systems, communications systems, RAD systems and for the...

Follow Us
The opinions expressed above are the personal opinions of the authors, not of HP. By using this site, you accept the Terms of Use and Rules of Participation.