While the benefits of both hybrid cloud and public cloud are well-documented, both introduce unique challenges for IT operations teams. One of the most potent challenges is application performance monitoring (APM).
For the most part, organizations have managed to corral their on-premises application performance. Using application monitoring tools ranging from SAP's Solution Manager to infrastructure monitoring platforms from vendors including Riverbed Technology, Dynatrace and others, enterprises have a pulse on traditional application performance. Yet, as companies migrate applications to the cloud or split resources between on-premises infrastructure and cloud infrastructure, monitoring becomes more of a challenge.
Keeping tabs on applications more than log monitoring
Application performance monitoring requires more than the ability to review system logs or to peruse stale data from agent-based products. Operations teams need to work hand-and-hand with developers to identify opportunities to dynamically adjust system resources. The best options offer a proactive approach. In the enterprise virtual infrastructure space, for example, VMTurbo uses performance data to adjust system settings and virtual machine (VM) placement. Operations teams can benefit from a similar approach when considering the needs of hybrid applications.
The distributed nature of hybrid applications means distributed management domains. Take, for example, a hybrid application that leverages on-premises compute for transactions and cloud-based VMs for data analysis. In this case, just as they would for legacy applications, APM tools must account for the performance of storage, compute and networking. But because the VMs could be anywhere, enterprise operation teams don't have access to the underlying infrastructures. This limitation means hardware-based monitors are not an option.
Netflix Chaos Monkey project shows what can be done
The most successful approach to track hybrid performance is to actually integrate application monitoring tools within the application. Netflix's Chaos Monkey performance monitoring project, which focuses on the application layer, is a great example. The service, which runs in the Amazon Web Services environment, automatically allocates resources to meet demand. If the load increases, resources are released; conversely, if demand subsides, resources are spun down.
Infrastructure as a service isn't the only consideration. Organizations are beginning to leverage software as a service (SaaS) and platform as a service (PaaS) approaches as well. Simply installing a management agent on a VM doesn't meet the needs of most organizations because there isn't an OS exposed to the customer on either PaaS or SaaS offerings.
New enterprise approaches are beginning to enter the market
Clearly, not every organization has the resources or the need to leverage an open source platform like Chaos Monkey to meet its performance monitoring needs. Filling that gap are new SaaS-only options from companies such as SignalFX. Last month, the startup, which is based in San Mateo, Calif., introduced hybrid infrastructure monitoring software that weds performance tracking and analytics. The result is an APM tool that allows customers to correlate performance and event data, which allows users to proactively adjust their hybrid infrastructures or take steps to negotiate new service-level agreements if necessary.
In the meantime, legacy traditional performance vendors, such as Splunk, are also beginning to fine-tune their offerings for the hybrid cloud. While Splunk doesn't provide real-time performance monitoring and analysis, the software does enable powerful event correlation.
What to expect when moving apps to the cloud
Using cloud APM for your benefit
When cloud apps fail