Nomad_Soul - Fotolia

Manage Learn to apply best practices and optimize your operations.

Mission-critical workloads not always cloud cost budget-buster

Controlling mission-critical application costs starts with knowing the expenses. With app resource requirements subject to change, knowing the numbers for monthly expenses can prevent budgeting bugaboos.

Every workload is going to demand cloud computing resources and services, which will pose a monthly expense to the business. It's important to understand the CPU, memory, disk and network I/O resources required by a mission-critical application, as well as how those resource demands will change over time.

Figuring out these numbers and cloud costs is usually accomplished by monitoring the mission-critical workload in-house during normal operations, and then correlating its performance to resources. The goal is to select a cloud instance or service type that offers adequate resources to meet the workload's peak requirements, but also minimizes the waste of idled resources.

For example, a large general-purpose Amazon Elastic Compute Cloud (EC2) instance with 2 vCPUs, 7.5 GB of memory and 32 GB of solid-state drive (SSD) storage space is priced at $0.14 per hour. But if there are situations when the workload might need more vCPUs, more memory or more storage, an extra-large EC2 instance with 4 vCPUs, 15 GB of memory and two 40 GB SSD disks may be required -- and that costs double, at $0.28 per hour.

The most difficult workloads for cloud deployment are those with erratic, unpredictable or rapidly growing resource needs.

Resource shortages can have a profound impact on workload performance, and they lead to larger resource allocations, which are costlier for the business. Part of your initial evaluation of a cloud provider should be to weigh the impact of resource shortages on mission-critical workload performance and understand the process and costs -- if any -- needed to change instances on demand or make other adjustments to resource allocations. For example, if it takes too long to change instances, or the larger instances become cost-prohibitive for the business, it may make sense to forego the cloud and leave the mission-critical workload in-house.

The most difficult workloads for cloud deployments are those with erratic, unpredictable or rapidly growing resource needs. Such workloads may be better accommodated in-house where the business can exert direct granular control over resource allocation to the virtual machine.

About the author:
Stephen J. Bigelow is the senior technology editor of the Data Center and Virtualization Media Group. He can be reached at

Next Steps

Looking into cloud cost analysis tools

Lowering cloud costs hampered by IT question marks

How to track spending with multiple cloud providers

Dig Deeper on Cloud pricing and cost optimization

Join the conversation

1 comment

Send me notifications when other members comment.

Please create a username to comment.

This article felt a little 'hand wavy' to me.  How many companies really spend the performance testing and engineering time to actually figure out how their system performs enough to really be able to predict costs?  I have a feeling the number is not as large as it should be.