BACKGROUND IMAGE: eugenesergeev/iStock

E-Handbook:

Smart strategies for application monitoring in the cloud

Manage Learn to apply best practices and optimize your operations.

5 tips to right-size your cloud instances and VMs

Improperly sized cloud instances and VMs can wreak havoc on your budget and decrease performance. Delve deeper into right-sizing with these five best practices.

Enterprises make it a top goal to optimize their cloud environments for cost and performance -- poorly sized instances and VMs can hit budgets hard and negatively impact user experience.

IT teams utilize the elasticity of the cloud to acquire, change and remove resources as needed. However, it isn't unusual for cloud-based loads to fluctuate, which means that a once-properly sized instance is suddenly too big or too small.

Follow these five tips to learn how to right-size cloud instances and to recognize the warning signs of a poorly sized VM.

Right-size a cloud VM for better performance, lower cost

Wrongly sized instances are one of the major contributors to wasted cloud spend. Ideally, users should right-size their machine images in the development phase. Start by defining the elements and features an application needs to run properly -- such as tool requirements -- and discard anything that isn't needed. Users should consider adjusting configuration parameters and VM memory size, since memory resources are commonly over-allocated.

Finding the right-sized VM also involves choosing the right instance type. Users can choose between reserved, on-demand and pre-emptible instance types. Reserved instances -- such as AWS Reserved Instance and Azure Reserved VM Instance -- are available at all times and best fit applications that run constantly. On-demand instances are a good fit for most apps and are free from long-term contracts, but there is still a risk of delay. The only way to ensure you made the right decision is through testing, which can help you find the best cost-performance balance.

Don't overprovision public cloud capacity

A traditional enterprise IT team must alter its mindset when it moves to the cloud to ensure its users don't overestimate resource use. With capital-intensive, on-premises deployments, IT teams spend considerable time determining their compute needs ahead of time so they don't get stuck with the wrong machine size. In the cloud, organizations can change, test and resize instances at will, but they also need to constantly monitor and evaluate usage to thwart overspending.

On-premises applications typically scale vertically, which give admins little choice but to reboot onto a different sized instance when capacity needs change. This process causes a service disruption, which can be costly and can negatively impact user experience. On the other hand, cloud-based horizontal scaling enables admins to deploy multiple instances at once to handle excess load. Organizations can also use autoscaling to give users performance and availability flexibility, without any fears of excess cloud capacity or wasteful spending.

Testing is a key part of right-sizing cloud instances and lets users know whether to redesign or reconfigure the application. Cloud instance sizes aren't fixed, but enterprises should keep a tight grip on who is authorized to control operations. With fewer authorized users, admins will get a clearer sense for where spending and usage occurs.

What are some warning signs I need bigger cloud computing instances?

It's not always possible to choose the perfect instance size on the first shot.

Various warning signs point to a poor fit, such as long runtimes, inability to respond to increased demand and having to add more instances to support a workload. Cloud usage reports can help IT teams determine what they need to add or remove, such as virtual CPU cores, storage and memory. 

Admins will often downsize cloud instances  to save money, but this can backfire if it's taken too far. Applications have certain baseline requirements they need to run smoothly. If the app's resources drop below those minimums, workloads will spend considerable resources overcoming bottlenecks. That's why it's crucial for IT teams to find the sweet spot in terms of instance size.

A custom cloud instance lets admins get choosey

All major public cloud providers offer different off-the-shelf cloud instances. Although these standard instances are useful in many cases, they won't work for all workloads and organizations. Instead, enterprises may find custom instances better suit their needs.

A custom instance enables users to pick and choose specific memory, CPU and storage resources to fit specific application requirement. For example, an application might be CPU intensive, but it doesn't require much memory or storage. Enterprises can save money by customizing an instance with fewer resources. AWS, Google and Microsoft all offer the ability to customize and build instances to match the exact needs of the user's workloads.

The more IT teams know about their applications, the better chance they have of obtaining a properly-sized cloud instance. Also, new paradigms frequently emerge, so admins should stay on top of the latest cloud platform updates to ensure their workloads utilize the best available tools.

Cloud optimization hinges on smart instance size selection

Enterprises have scores of cloud instance types and sizes to choose from -- perhaps too many. The instance you choose should meet all application performance requirements, but these requirements can vary. This makes it a challenge to determine the proper size for a cloud instance.

To understand all available options, users should check their cloud provider's resource limits for RAM, CPU and storage. Utilization metrics can show if the instance is right for the specific application. If there are usage spikes, users should find an instance that has ample space to handle the surges.

Enterprises should also consider dynamic cloud instance types to boost performance, such as AWS burstable instances. Load tests, which should be models from real-life usage patterns, can help predict performance metrics and determine which instance type and size to select.

Dig Deeper on Cloud application monitoring and performance

Join the conversation

1 comment

Send me notifications when other members comment.

Please create a username to comment.

What techniques have helped you right-size tricky cloud workloads?
Cancel

-ADS BY GOOGLE

SearchServerVirtualization

SearchVMware

SearchVirtualDesktop

SearchAWS

SearchDataCenter

SearchWindowsServer

Close