As organizations rushed to embrace digital business transformation initiatives in the last year in response to rapidly changing economic conditions, the number of mission-critical applications that now need to be managed has increased significantly. A large percentage of those applications are based on a microservices architecture that, while being more resilient and flexible than legacy monolithic applications, are a lot more difficult to manage.
Each microservice has dependencies on other services that it invoked over an application programming interface (API). Those dependencies span the underlying infrastructure, other microservices, and monolithic applications that may be running on the same cloud, in another cloud, or in an on-premises IT environment.All of those dependencies are forcing IT organizations to revisit how they monitor those increasingly complex environments.
Historically, IT organizations employed an array of tools to monitor each IT platform employed. When there is an issue, IT teams would gather in “war rooms” to compare monitoring in the hopes of eventually identifying the root cause of an issue. Today it’s apparent IT teams require tools that provide a lot more context and visibility into their IT environments otherwise known as observability.
“It’s not just monitoring renamed,” says Joe Fitzgerald, vice president and general manager of the Management Business Unit at Red Hat. “You can’t manage what you can’t see.”
Monitoring Infrastructures with APMs
Most application performance management (APM) platforms can to some degree now also monitor infrastructure. Providers of tools that historically only monitored infrastructure have conversely been adding the ability to monitor applications. This shift is at the core of both an acquisition of AppDynamics, a provider of an APM platform, by Cisco way back in 2017 as well as the more recent acquisition of Instana, a provider of an APM platform optimized for microservices, late last year by IBM.
As it becomes simpler to observe applications using integrated platforms it also theoretically affords IT teams the opportunity to consolidate many of the tools they have licensed over the years. Eliminating those license fees helps in part justify the return on investment in what are now being touted at next-generation observability platforms.
However, it still requires a lot of time and effort to instrument applications. IT teams still need to find a way to insert agent software within their applications that collects all the logs, metrics, and tracing data needed to truly observe an application. Deploying and maintaining all of that agent software is why many IT teams have tended to limit their usage of APM platforms to only their most mission-critical monolithic applications.
A second major factor is the cost of the APM platform itself. Budgetary constraints would limit how broadly APM platforms could be employed across an entire application portfolio.
Now there is an effort to reduce the cost of instrumentation using open source agent software. The Cloud Native Computing Foundation, for example, has launched an Open Telemetry project that provides developers with the ability to instrument their applications. Most providers of APM platforms have signaled their intent to support these open source initiatives as an alternative to proprietary agent software they have developed. However, most of them will also note their proprietary agent software, at this point at least, collects a lot more data and is generally easier to install.
Also read: Best IT Management Tools for 2021
SaaS vs. Open Source Platforms
Regardless of type of agent software employed organizations will not be able to manage microservices-based applications without including some instrumentation capabilities. Microservices-based applications are designed to degrade gracefully by rerouting applications calls rather than outright crash in the event of one or more components of an application suddenly becoming unavailable. The downside is in the absence of any instrumentation capabilities it might take weeks to determine what the root cause of performance degradation actually is. In effect, IT teams that employ microservices-based applications to drive digital business transformation initiatives will be required to embed agent software within those applications to enable observability.
The issue then becomes to what degree they may want to rely on a software-as-a-service (SaaS) platform to provide that observability versus deploying a commercial or open source platform in cloud or on-premises IT environments themselves. Prometheus, an open source monitoring tool developed under the auspices of the CNCF, for example, has gained traction among developers deploying applications on Kubernetes clusters.
Providers of SaaS-based platforms are betting most IT teams, however, would prefer to make use of a SaaS platform that is maintained by a vendor. In fact, AppDynamics just announced it is leveraging availability zones on the Amazon Web Services (AWS) cloud to add additional compute horsepower in Singapore, London, Hong Kong and, later this year, in São Paulo and Cape Town.
As IT environments become more distributed it’s become more critical to be able to surface insights closer to the data center environments where applications are deployed, notes Vipul Shah, chief product officer for AppDynamics. In fact, consumption of the AppDynamics platform today is greater than ever, adds Shah. “We’re seeing peak utilization not seen prior to the pandemic.”
Also read: It’s Now or Never for Shift to Real-Time Apps
AI-powered Observability
Ultimately, the rise of artificial intelligence (AI) capabilities will most likely force a shift to the cloud. Most individual IT teams are not going to be able collect enough data on their own to build and then maintain an AI model optimized for observability. In contrast, providers of observability platforms will pervasively embed AI capabilities in every offering, says Mughees Minhas, vice president of product management for enterprise and cloud manageability. “There will be AI embedded in all observability and management platforms,” he says.
IT organizations that need to observe applications that are at the core of a strategic bet an organization is making on a digital transformation initiative will need to determine to what degree are they really competent enough to build, maintain, and secure a true observability platform that can provide similar AI capabilities. Chances are good the number of internal IT teams that can will be few and far between. And, even then, the business is not likely to have much patience for IT teams that may still want to devote resources to rolling their own observability platform.
Read next: Open Source Platforms Vie with IT Vendors for Management of MLOps
