Jim Manias is Vice President at Advanced Systems Concepts, Inc. and is responsible for the overall market strategy and planning for a range of products.
Flexibility. Scalability. Accessibility. These are just some of the many benefits organizations are achieving with virtual and cloud computing resources. Widespread cloud adoption is shifting how companies use resources so that they can scale quickly to meet changing business needs. Hybrid solutions are becoming more common as companies still retain some on-premise functions while moving much of their data and processes to the cloud. However, despite the benefits of the cloud, managing and monitoring the spin up and spin down of these systems poses a significant challenge for the IT organization.
One issue developing is the use of legacy infrastructures with cloud computing environments. Data center operators and IT are seeing problems arise where hardware-based environments take a “bottom up” approach to policies, and cloud computing takes a “top down” application-centric approach. The delivery of services can be interrupted when older technologies are used with the cloud. Utilization rates and the quality of service can also suffer.
This issue is aggravated by companies that apply the same management principles to virtual machines (VMs) as they do to physical hardware, resulting in inefficient sprawl. The phenomenon of “virtual machine sprawl” is a costly but hidden problem because the actual machines and costs aren’t readily seen (unlike an on-premises site that runs out of rack space). This type of sprawl occurs when VMs are left idle and unused, which results in concealed costs such as the consumption of storage and network resources.
There are considerable hardware costs for virtual hosts, which can eclipse single-use physical servers because they must be available for immediate scaling. They have massive amounts of RAM, network adapters, storage, and multi-core CPUs, all of which are expensive. Virtualization does drop the cost by moving up the density of servers, but cloud resources are finite and must be managed efficiently. Oftentimes, organizations forget these unused cloud resources are still running and incurring expenses.
Moving to a cloud environment requires a revised cloud infrastructure plan. Regulations, fast-moving business environments, and competition mean IT is at the center of the business and must manage resources and efficiency. Here are some tips for data center managers to keep processes moving smoothly in the age of “always on” processing:
- Limit the number of people that have access to create VMs. This simple step helps channel VM formal requests to select people with approval powers, causing IT staff to carefully consider if the request is needed, instead of allowing them to complete it with just a few clicks.
- Allocate a certain amount of resources to different departments using resource pools. You can control VM creation better if you actively limit the resources available per host.
- Monitor VM lifecycles so you can spin them up and down quickly to save money and better utilize computing resources.
For the data center, manual management of computing resources takes a considerable amount of time. Relying on individuals remembering to spin down systems results in an elemental approach that lacks governance, visibility, and scalability as the organization grows. A better approach is to put in place a sophisticated workload automation solution that will help you create guidelines for both virtual and physical machines.
Virtual machine sprawl is similar to running the clothes dryer when the clothes are already dry or setting an office’s AC to 65 degrees over a holiday weekend. The result is a waste of resources and a jump in the electricity bill. Similar to a smart thermostat, some of the more advanced workload automation solutions utilize a smart queue to automate the spin up and down of systems. When additional computing resources are needed to meet an SLA or an unexpected business demand, smart queue will spin up additional VMs to help the job or plan be completed in the allotted time frame. Similarly, smart queue will spin down idle systems that are running but not being utilized.
Reducing this sprawl is best combatted by improving how the data center manages resources in the cloud. With workload automation, organizations can take an architectural approach to cloud resource management. This means building a framework for the spin up/spin down of systems that provides greater governance and control over all of the organization’s virtual and cloud assets. As a result, organizations can better meet SLAs, reduce costs associated with manual management, and increase IT agility.
Industry Perspectives is a content channel at Data Center Knowledge highlighting thought leadership in the data center arena. See our guidelines and submission process for information on participating. View previously published Industry Perspectives in our Knowledge Library.