Observability tools are broadening the scope of data they gather and deepening their analysis of how app developers' updates affect IT performance as enterprise DevOps practices mature.
Recent vendor acquisitions and product updates, such as LogicMonitor's acquisition of Airbrake and Dynatrace's integration of Keptn into its Cloud Automation Module, take different stances on IT automation, but are similarly focused on developers and integration into DevOps pipeline workflows.
These moves come as the Agile approach to software delivery matures within enterprises, leading to frequent, iterative changes to apps via CI/CD pipelines, often distributed into microservices and deployed within complex cloud-native infrastructures such as Kubernetes clusters.
These trends, in turn, have driven a shift towards observability, an evolution of IT monitoring in which systems flexibly query a broad volume of data to get to the root of IT issues. Observability tools for production monitoring have also matured over the last year, including industry standards such as OpenTelemetry.
Now, IT organizations are beginning to tie these two domains together, weaving observability tools more tightly into software delivery pipelines, prompting vendors to update products accordingly.
"Everyone is trying to get a full suite of tools in place, and get closer to the developer," said James Governor, analyst and co-founder at RedMonk. "Better troubleshooting tools for engineering teams … is arguably the key aspect of observability."
LogicMonitor, previously an infrastructure-focused monitoring vendor, will use Airbrake's application performance monitoring and code error monitoring features to enhance an APM tool it plans to ship by the end of 2021. Airbrake's software connects application performance data to software bugs and integrates with developer workflow tools such as Atlassian Jira, GitHub and Slack.
Airbrake also covers more than 20 application programming languages, can monitor mobile apps and will contribute to LogicMonitor's plans to offer its customers centralized visibility into the full IT stack from application code down to individual network devices, said Tejaswi Redkar, Chief Product and Technology Officer at LogicMonitor.
James GovernorAnalyst and co-founder, RedMonk
"Ultimately developers want to go and fix bugs, and this information is extremely valuable, [showing] in which part of code an error is happening, that [developers] can quickly fix before it escalates," Redkar said. "What LogicMonitor brings to the table is monitoring of the underlying infrastructure … so it brings a different persona who is also responsible for the health of the application along with the developer into one contextual pane of glass."
The fabled "single pane of glass" has been part of IT vendor messaging for over a decade, with varied results, but achieving that goal is becoming more critical as IT infrastructures grow more complex and app changes occur faster and more frequently, IT pros say.
"It's important for LogicMonitor to bring these features into their stack to keep up with modern technology patterns," said Andy Domeier, senior director of technology operations at SPS Commerce, a Minneapolis-based communications network for supply chain and logistics businesses and a LogicMonitor customer.
"There is more segmentation in our technology stacks, but the interdependencies between [those different segments] are critical," Domeier added. "The better we can instrument all the layers and understand the performance of them, the better we can deliver services [for] customers."
Dynatrace Keptn reaches deep into DevOps pipelines
Dynatrace, which began as an APM vendor, also broadened its data-gathering to encompass multiple layers of app and IT infrastructure over the last two years, including Kubernetes cluster metrics, system logs, and most recently, native log data formats. In addition, the observability vendor began to expand beyond monitoring into DevOps pipeline automation with the Cloud Automation Module it introduced Feb. 10.
The Cloud Automation Module includes the vendor's first supported version of its Keptn open source project, launched in 2019. Keptn integrates with DevOps pipeline tools such as Jenkins and GitLab, as well as GitOps tools such as ArgoCD, to automate application lifecycle processes in response to app performance data. IT pros create Service Level Objective (SLO) quality gates using Keptn, and Keptn automatically fails or promotes builds within DevOps pipelines according to whether they meet quality gate criteria.
The open source Keptn project is already used in production by enterprises such as Intuit and clinical data management company ERT, according to presentations at a virtual event earlier this month.
"Before implementing Keptn … there were a lot of things which were manual [in a Kubernetes environment where] we wanted to do testing on 1,000 nodes and 200 to 300 namespaces simultaneously," said Sumit Nagal, principal engineer at Intuit, in a presentation Feb 10. "Self-service was completely missing, especially for SLO quality gates."
Intuit built test automation workflows using Keptn and Gatling open source tools into its ArgoCD GitOps pipeline, according to Nagal's presentation.
"We set up Keptn on one of our Kubernetes clusters," he said. "Now there's no one who needs to [manually] work on [testing] -- it's all automated as part of the pipeline."
The commercial Cloud Automation Module for Dynatrace's Software Intelligence Management platform, which will become generally available over the next 90 days, adds packaged integrations with CI/CD pipeline tools, as well as release inventory, version awareness, and automated reporting and alerting features through the Dynatrace UI.
Dynatrace Cloud Automation Module and Keptn take IT automation a step beyond LogicMonitor's current plans for Airbrake, which doesn't directly orchestrate DevOps pipelines.
In general, however, IT pros have begun to demand tighter integration between observability tools and developer workflows as they "move fast and break things" under Agile/DevOps paradigms, RedMonk's Governor said.
"We're making more changes [to application code] than we ever were before, and even though we're doing testing before we deploy, things are going to break," he said. "We're dealing with non-deterministic, distributed, complicated environments, and when they break, we want to be able to identify what changes may have caused those breakages as quickly as possible."