4 ways to stabilize virtualized application stacks
- By Joel Dolisy
- Dec 18, 2015
Building an application stack in today’s dynamic government IT infrastructure is like creating a virtualized house of cards. When this house falls, however, so does application performance. That’s because everything that goes into an application stack -- from the applications themselves to the tools used to manage them to the operating systems they run on -- are all highly dependent upon one another. When one fails, they could all falter.
The changing state of IT makes this house of cards even more precarious. Things are constantly moving and shifting. New applications, systems and solutions are being plugged in here and there, while old technology is being replaced by new advances.
However, even in today’s increasingly unstable application stack world, there are a few things you can do to ensure your foundation remains firm and stable.
Start strong when starting from scratch
At the start, make sure you have a good understanding of the priorities for applications and their operational characteristics. For example, you may have to prioritize applications with high CPU or memory requirements. In any case, get a good lay of the land in order to allocate your resources appropriately.
Next, properly align all of those application resources, including database and storage layers. If this step is not followed, it’s going to be very hard to maintain optimal application performance.
When working with an existing application environment, first get a good handle on baseline performance and ascertain how potential changes might impact that performance. Also, take this opportunity to estimate the cost vs. benefits of these potential changes, to see if they’re even worth the effort.
Gain as much visibility as possible
Understanding baseline performance involves gaining visibility into the application stack’s overall performance. Without visibility, you will not be able to see and address problems as they develop. At that point, it doesn’t matter how solid the application stack is, because there’s a good chance it will all fall down.
Virtualization monitoring tools can deliver a great deal of insight into the performance of an application stack, providing immediate warnings about potential problems and allowing quicker response. They can help pinpoint specific application issues, so you’re not left hunting and pecking through application stacks trying to find causes of problems.
Actively and continuously monitor existing resources
The important process of allocating resources must be ongoing to accommodate dynamic changes in the application environment. Things change over time; a large resource pool that’s very active today may not be tomorrow, for example. It’s critical to check resource pools frequently and adjust them as necessary to effectively manage activity and maintain optimal performance.
An application monitoring solution can help with resource allocation needs by providing insight into appropriate allocation. It can also deliver information pertaining to the availability and overall performance of all of the network’s applications and how they can impact the infrastructure, databases and more.
Consider different approaches to virtual machines and application management
When it comes to grouping and managing VMs and applications, there are two possible approaches.
The first is to group high-demand applications on a single host or data store and allocate plenty of resources to their performance. This can help maintain the most important applications.
Alternatively, applications can be mixed on a VM. This prevents a single resource from being inundated with a flood of CPU or memory demands at one time. You might have to kill an entire VM if there’s a critical app failure, but this could be bette r than trying to sort through the mix to try and find the problem.
That last point pretty much sums up the overall challenge with application stacks: they’re extremely delicate and intertwined. This makes them difficult -- though certainly not impossible -- to manage. You just need to be sure to take a measured approach -- and place your cards very, very carefully.
Joel Dolisy is the CIO at SolarWinds.