OpenTelemetry is hitting its stride as a powerful alternative to vendor lock-in and patchwork services.
An open-source observability framework for creating and collecting telemetry data, OpenTelemetry is a Cloud Native Computing Foundation (CNCF) project that ranks as the second most popular (behind Kubernetes) for the organization and counts more than 800 monthly active developers as contributors.
Telemetry is the signals, traces, metrics, logs and other types of data emitted by applications and their infrastructure that developers and DevOps engineers want to capture and then analyze to reap the benefits.
OpenTelemetry or OTel takes telemetry a step further by providing a single, vendor-neutral, unified standard to measure and express performance data. It can capture data effectively from any source running in any environment, including on premises and in the cloud.
Key supporters of the CNCF project include heavy hitters such as Google Cloud, Amazon Web Services, Datadog, Microsoft Azure and Splunk.
“Last year would be the year when suddenly hundreds, if not thousands, of organizations who previously were not that deep into the weeds in this space said, ‘Oh, this OpenTelemetry thing, this makes complete sense. This gives me a ton of value. We're going to switch to it,’” said Morgan McLean, director of product management at Splunk and a cofounder of OpenTelemetry.
Targeting the vendor lock-in problem
Developers previously had two options to monitor their systems, according to Austin Parker, head of developer relations at Lightstep and the community manager for OpenTelemetry. They could cobble together the different system signals in incompatible ways, where one library would give logging information, another would give time-series metric data and a third might give tracing data – but they didn’t work together, Parker explained.
“I have all this different performance data, but I'm getting it all in different ways,” he said. “I'm getting it in different formats, and I'm using different tools to analyze it, and then you are responsible for stitching all that stuff together and making sense out of it.”
The second path would be a proprietary solution that provided all the different types of data, but left users locked in to a commercial vendor.
“So if they don't work with your particular system, your particular setup, your particular cloud, whatever you're trying to do, now you aren't able to extend it,” Parker said.
And it is cost-prohibitive to switch, McLean added.
“It doesn't matter which vendor you picked, they all had their own agent, and you were stuck with it,” he said.
Enter OpenTelemetry
The advantage of OpenTelemetry is that it “obliviates” both of those problems with an open-source solution that can be integrated into different clouds, services and systems without being tied to one particular commercial or even open-source solution, according to Parker. It takes all of the data, the instrumentation, the collection, the format of data itself and gives a single answer to apply.
“It gives a lingua franca for expressing that telemetry data and making it portable and useful, regardless of how you're analyzing it, how you're viewing it, what questions you're trying to ask of it,” he said. “We've taken this thing that used to be … two bad answers and given you one good consistent answer. You can choose what to do with the data later. You're future-proofing yourself in a lot of ways.”
OpenTelemetry also brings a lot of raw capabilities previously not offered, according to McLean. In addition to substantially better data capture, it also allows for a breadth of integrations with various technologies, whether those be programming languages for application performance monitoring, data collection, operating systems for host information or other third-party applications like databases, he added.
OpenTelemetry is in widespread use at major tech companies, large vendors such as Microsoft and Google, companies such as eBay and Shopify, and major financial institutions and telecoms.
On the bleeding edge, developers and their companies are using the OpenTelemetry data to understand how end-users are experiencing their products or services – whether it’s customers logging in and using their ecommerce site or banking through their app, for example. They can combine the performance insights that OpenTelemetry provides with their business goals to optimize around customer experience, Parker said.
The future is open
As the project grows in scope, contributors are adding OpenTelemetry features including new signals and new types of telemetry such as profiling.
“I think 2023 is going to be a really exciting year, where we're expecting to really have the major signals of metrics, logging and tracing all 1.0 in our core languages,” Parker said. “[Having] 1.0 for a lot of our core tooling is a goal for the year.”
The eventual goal is to make OpenTelemetry a built-in feature.
“Right now, there's kind of a choice you have to make,” Parker said. “You have to actively try to integrate this into your software. Over the next several years, we hope to kind of push it down a level and make this part of application runtimes and container runtimes and something that's just a built-in part of what you're doing as a developer.”
Are you using or working with OpenTelemetry? We want to talk to you. Send us a letter here.