As enterprises transition away from traditional architectures that host applications in physical servers within on-premises data centers to hybrid environments driven by public and private cloud assets, adequate application performance monitoring becomes a challenge. "The shift to SaaS is probably the leading agent of change in application adoption, and primarily because companies are seeking better flexibility, speed and agility," said Nik Koutsoukos, senior director of product marketing for Riverbed SteelCentral at Riverbed Technology. Finding that balance between control and agility is key as organizations continue to migrate away from old-school architectures. It's also something IT executives often struggle with as they weigh the benefits of SaaS against the loss of visibility that occurs when applications move outside the enterprise's perimeter.
Meanwhile, although reliance on the cloud may be increasing, it's rare for enterprises to abandon their existing architectures entirely. "Some are still dependent on pieces that exist behind the firewall," said Jason Lieblich, founder of Exoprise. In addition, he believes another part of the challenge comes from the notion of instrumentation versus ownership. "If you own all of those pieces, whether it's public, private or hybrid, if you own and take the time to instrument them, then you can get the telemetry into all those pieces." However, if an enterprise doesn't have ownership of all the various parts and pieces, then that approach becomes problematic. The process can also be time consuming and expensive. Seeing across all the environments becomes more difficult as a result.
This need to monitor and measure applications in distributed, heterogenous environments is something that isn't always well addressed in existing application performance monitoring (APM) platforms. "When these applications are running in a distributed fashion they're no longer monolithic, and people need visibility across them," explained Maneesh Joshi, senior director and head of product marketing and strategy at AppDynamics. "Legacy vendors aren't able to handle traffic that's traversing from public cloud to private cloud to private data center." Even as administrators are able to move applications or the assets they rely on from one environment to another, ensuring their APM solutions are able to follow suit becomes a hurdle all its own.
Find the right APM platform
A number of existing application performance monitoring solutions already have the right feature sets to support a move to the cloud. However, administrators have expressed some ongoing frustrations as platforms catch up to evolving environments. Dashboards, for example, may not provide the kind of data enterprises need, leading users to "think of them as static jpegs," Lieblich said. Rather than functioning as a meaningful representation of current performance metrics, he said, "no one really knows the science behind when they update." Administrators may be able to look back over time for performance trends, but identifying problems in real time remains a high priority for enterprises interested in gleaning the best efficiency from their networks. For this reason, real-time visibility into application performance is critical.
The ability to quickly respond to performance issues and impending needs is important for more reasons than just fixing problems after the fact. As the holidays approach and traffic on retail sites and their corresponding inventory and shipping partners ramps up, Joshi said APM solutions in hybrid cloud environments can help administrators "spin up VMs or more servers in-house, or even scale up the Amazon infrastructure their public cloud is running on." Because IT's role revolves around strategy as much as problem-solving, this ability to proactively address anticipated usage fluctuations is something enterprises should consider as they're shopping for cross-environment APM platforms.
If an enterprise anticipates retaining some of its legacy architecture, an application performance monitoring solution that can support monitoring in a traditional environment along with the new cloud components is a must. Koutsoukos said this includes the ability to examine deep application instrumentation or "tap into a physical network and pull packets…and analyze that data." Working with the cloud vendor may be a useful way to extend monitoring even after the application has left the on-premises infrastructure. "In order to do that, you need to be looking at technologies that have the ability to instrument into the cloud, and partner with the SaaS providers to give you that level of instrumentation."
Don't let cross-environment APM become a time sink
Even as administrators grapple with ways to gain better visibility into how applications are performing in their various environments, there's also a concern that the function of APM itself takes time. Making the task more complex as architectures expand to include public and private cloud raises worries about carving out enough time to manage APM correctly. The good news is that the old way of doing things is evolving, bringing time investments in application performance monitoring down considerably. "Today's technologies are much faster in terms of time to value," Joshi explained. Deployment timeframes of less than a week aren't uncommon, even when dealing with big or complex networks. "It doesn't matter how large the infrastructure is…they can start monitoring applications in production environments in under a week."
The idea of shifting things off-premises isn't much of an "if" question any longer. It's really about when. Koutsoukos said that finding a good solution early will help save time. "[Administrators] are going to be under pressure to move more and more things to cloud," he explained. "They very much need to bear in mind they'll need to have tools in place to monitor those applications irrespective of the fact that they are now not tied to a physical environment." The decision to move to a SaaS model or to adopt a hybrid architecture needs to include a commitment to deploying application performance monitoring capabilities that can encompass all of those environments. "Make the decision in parallel, and ensure you'll be able to have an adequate level of visibility when moving to a new architecture," Koutsoukos stressed.
Platforms with features designed for easy deployment, such as the inclusion of built-in sensors for common applications, may help enterprises with lean resources stay in the APM game, even as resources move to the cloud. "Having those out of the box, pre-trained and preconfigured, makes the deployment and the selection that much quicker," Lieblich said. Knowing where the organization's blind spots are likely to be once the move to the cloud begins—and taking steps to facilitate visibility into those areas before making the transition—will ensure that the organization is able to maintain performance and user experience no matter where applications reside.
Photo courtesy of Shutterstock.