Enterprises are inundated by a plethora of applications. This, coupled with an exponential growth in data generated within the enterprise, has led to an increase in the demand for computing resources.

This demand has not been not a major concern for data center managers until recently, when energy prices and data center hosting costs started to rise considerably. To add to these challenges, data center managers discovered that most of their systems were using only 10 to 15% of allocated capacity.

Virtualization has helped by using a single machine to host multiple virtual machines, thereby increasing overall infrastructure utilisation and offering substantial cost savings to IT.

Virtualization also brought in other benefits such as ease of management, faster provisioning/de-provisioning, dynamic allocation of resources to virtual servers, highly resilient infrastructure and agility to business IT services.

One of the other significant benefits that virtualization brought in was enablement of cloud-based IT services.

Virtualization enabled various forms of cloud services - private and hybrid cloud. Each form of cloud service offers a different set of benefits and challenges.

However, “cloud sprawl” is prevalent across all forms of cloud services and it is being identified by most IT managers as a critical challenge.

Flexibility at a price

Cloud sprawl is defined as an “uncontrolled growth of computing resources underlying cloud-based IT services that exceed the resources required for a definite number of authentic users”.

This sprawl can be a result of unmanaged virtual machines (VMs) — that is, VMs no longer required or VMs created without consent of the corporate IT team. Rapid proliferation of VMs has serious implications with regards to data center space, support and maintenance, energy consumption and various licensing issues.

It is mainly the associated cost implication that worries most IT managers and CIOs. A typical private cloud service is made up of application software, compute resources (memory and processor), storage memory, network connections, and network devices. To calculate the true cost of cloud, the cost for all the above components needs to be taken into account.

The entire tech needed for networking/management/dynamic workload balancing and hosting an enterprise cloud platform can cost anywhere between US$25,000 and US$35,000, and can support a consolidation ratio of 6:1.

The cost of an individual virtual machine in such a scenario would be approximately US$4,500. The cost of a cloud sprawl would be proportional to the cost of unused resources as mentioned above.

These costs would further increase depending on the amount of resources, which would include staff and computing systems, used to support these unused resources.

Cloud sprawl management

Controlled cloud sprawl is a sign of healthy cloud management and can be achieved by using good practices in IT operations.  Organizations can use the following as a checklist when  developing a detailed strategy to control cloud sprawl:

Define and communicate the cloud strategy clearly:

The cloud strategy, related processes, benefits and costs need to be communicated clearly to the user community and appropriate expectations should be defined.

Audit the cloud services periodically:

It is important to perform regular enterprise cloud services heath checks. These internal audits need to cover all aspects of cloud services including applications, the virtualization platform itself and underlying infrastructure components. This will also help organizations ensure that cloud services are aligned to business needs at all times.

Build transparency in enterprise cloud-related processes and systems:
Lack of visibility can often have an impact on user perception of services and the cost of making those services available. Enterprises need to adopt industry good practices around service catalogue, request fulfilment, and demand and capacity management, and integrate them to present streamlined / transparent cloud management processes to business users.

Build a dashboard for monitoring cloud services performance:
Dashboards are useful to monitor service performance. A cloud services dashboard for IT management that shows performance of various aspects of cloud services in a single view will help management to identify opportunities for improvement earlier and avoid additional costs that might result from delayed decision making.

As it becomes easier for non-IT users to provision services in the cloud, a strong governance framework for provisioning and consumption of cloud services will be crucial for keeping the cloud business case intact.

However, cloud sprawl can be used by organizations as a crucial parameter to gauge enterprise cloud health. By using the steps mentioned above, organizations can prepare a comprehensive plan to control unauthorized proliferation of computing resources.