What is Application Performance Monitoring in the Cloud Era?

Can you identify your application delay issues, feature uptake rates, and abandoned sessions? Do you know which parts of the Cloud are positively and negatively affecting your users right now?

Poor response time and poor usability are the top-two user complaints, but according to Aberdeen research, only 3% of organizations that consider performance management to be “just a development activity” can accurately identify the sources of their issues.

Performance management of applications today is not just about what is going on in the code, nor is that what customers care about. It does not matter to your users if network behavior, cloud workloads, or their device of choice is causing their problem. Users may abandon a session for a number of reasons: sub-optimal responsiveness, too many steps, or because they are still doing research. They may not use a new feature for multiple reasons, either; because it is slow, because of its placement on the screen, or because of how it performs on their preferred device.

If you are only looking at the performance of your code during the development cycle, you will be largely unaware of these additional elements.

APM is about experience

Quality of experience in the Cloud is the sum of multiple factors, not just application execution time, and it can vary significantly. Performance fluctuates due to network traffic, workloads that share hardware, and the behavior of any intermediate services, in addition to your own code.

In this complex environment, modern Application Performance Monitoring (APM) tools are quickly becoming a necessity. End-to-end transaction activity, based on deep visibility of all transactions — for all users, all the time — supplies the data you need to derive meaningful insights about user experience. Advanced analytics and artificial intelligence use the big data generated by APM to detect the source of performance anomalies and quickly point to the most important issues.

For example, Maersk Line, the world’s largest container shipping operator, has seen their revenue change from 20% digital sources to over 90% digital over the past five years. From the container-booking process to customers tracking their shipments or transacting business processes aboard ships, visibility and transparency into and through the Cloud are critical to the scalability and stability of their applications. Application discovery, user-experience monitoring, and detailed application granularity support the need for fast and appropriate problem remediation.

APM is about DevOps and sharing

DevOps refers to the modern software engineering practice of merging development, QA, and IT operations into an agile and responsive activity covering the entire application lifecycle, from initial code development to live application monitoring. The objective is closer alignment with the business, and the practice typically features short development cycles and frequent code changes in response to a dynamic environment.

The DevOps culture is extending APM to the broader team, sharing information with multiple stakeholders to improve collaboration and detect and resolve problems sooner.

Advanced APM systems support the DevOps culture with a variety of tools and techniques, such as:

• Automated code discovery captures activity without requiring manual instrumentation and includes a compare mode to see the before and after effects of changes.
• Unified dashboards and integration with common tools correlate logs to application data, automatically open tickets, and expedite the troubleshooting process.
• Performance graphs of user transactions and their financial implications enable business-unit team members to quickly prioritize the efforts for maximum impact.

APM and SLAs

Cloud providers and their software-as-service (SaaS) solutions are increasingly critical to DevOps, as they enable reduced costs and time of application development. However, they introduce additional challenges to APM, such as their impact on the end-user experience and the need to hold providers accountable to their service-level agreements (SLAs). With workloads spinning up and down on demand, shared resources moving between cloud instances and geographic locations, and SLAs with multiple characteristics, finger pointing between external providers and internal teams exacerbates troubleshooting.

Service levels include multiple characteristics, such as uptime, availability, bandwidth, I/O rates, and CPU utilization. Detailed APM reports linked to actual user experience metrics can make it much easier to monitor and enforce SLAs. Teams can quickly evaluate the business impact by different characteristics such as department, geography, device type, or operating system. In addition, organizations can define SLAs with providers based on user interactions, workflows, or business processes that are directly related to their own business (instead of the provider’s infrastructure).

Moving from proactive to pre-emptive

The cloud era is delivering tremendous opportunities for digital transformation of business processes and customer experiences, including increased agility, new devices, and reduced costs. Many of these gains have come from slicing the operating environment into smaller and smaller pieces and giving them greater freedom of movement.

APM in the cloud era gives organizations a comprehensive view of their users’ experiences in this new reality, spanning networks, applications, and devices. As Andy Laurence, Head of Production Services at Maersk Group Infrastructure Services noted, the net effect is a “circle of development that helps us to be responsive, pushes us to be proactive, [and] positions us to be predictive, which then allows us to be pre-emptive.”

Gayle Levin is director of solutions marketing at Riverbed Technologies. Previously, she held product marketing and campaign roles at VMware, Oracle, and Splunk as well as several startups. Her interests lie in the impact of technology on the way we think and work today.