Advantages and disadvantages with using OpenTelemetry over collecting data with the OneAgent

[Benedikt]

Hello all, 

I am currently debating about using OpenTelemetry to collect data and then send it to Dynatrace or collecting data with the OneAgent.

Have you ever made any experiences with this? 

My question is, if you think it is valuable to use OpenTelemetry instead of the OneAgent? What comes out-of-the-box with using OpenTelemetry and what is missing compared to the out-of-the-box functionalities from OneAgent? 

In general, what would you say are the advantages and disadvantages with using OpenTelemetry over collecting Data with the OneAgent? 

Greetings and thank you for your help, 

Benedikt

[radek_jasinski · Accepted solution]

Hi @Benedikt 
I was preparing such a comparison for my client with advantages and disadvantages against OA. I attach it here (sorry if something is not coherent, because I put this text in the automatic translator, as I have the document in Polish:))

Otel:

Advantages:
1. OpenTelemetry is an open-source project that follows open standards for telemetry data (traces, metrics, and logs). This can be beneficial for ensuring long-term compatibility and avoiding vendor lock-in.
2. Being open-source, OpenTelemetry allows for customization and extensibility. You can modify or extend it as needed to fit your specific use case.
3. OpenTelemetry supports a broader range of programming languages, which can be advantageous if your stack includes less common languages.
4. a community-driven project, it benefits from the input and development efforts of a diverse group of contributors.

Disadvantages:
1. OpenTelemetry might require more effort to set up and configure compared to vendor-specific solutions like OneAgent.
2. Depending on your existing tools and platforms, integrating OpenTelemetry might require additional work.
3. While OpenTelemetry captures a wide range of telemetry data, it might lack some advanced features specific to vendor tools like Dynatrace.

OA:

Advantages:
1. OneAgent is designed to be easy to install and configure, offering a more plug-and-play experience.
2. Dynatrace offers advanced features like AI-powered analysis, automated root cause detection, and deep-dive diagnostics, which may not be available with OpenTelemetry.
3. Being part of the Dynatrace platform, OneAgent offers seamless integration with other Dynatrace features and tools.
4. OneAgent provides automatic and real-time instrumentation with minimal configuration.

Disadvantages:
1. Using OneAgent ties you more closely to the Dynatrace ecosystem, which could be a consideration if you aim for flexibility or anticipate changing monitoring needs.
2. Dynatrace is a commercial product, and using OneAgent might incur more costs compared to an open-source solution like OpenTelemetry.

Conclusion
- You prioritize open standards, need support for a wide range of languages, or require custom telemetry solutions.
- You want a more straightforward setup, advanced AI-driven insights, and tight integration with the Dynatrace ecosystem.

Radek

[Benedikt]

Hello Radek, 

thank you very much! I briefly read through your post and I can say it is very well understandable.

That will help me a lot.

P.S. for others: other experiences would still help me, thank you 🙂 

[Slawa · Accepted solution]

Keep in mind consumption costs.
You will have benefits in case of low traffic and high RAM config on hosts
Consider one agent in case of high traffic.

------------------
Otel billing per trace(traffic based)
One agent billing is per GB (RAM config based)

[Benedikt]

Hello Slawa, 

thank you for your reply. Do you have any link providing the cost of Otel? I can't find anything about it.

[Slawa]

https://docs.dynatrace.com/docs/manage/subscriptions-and-licensing/monitoring-consumption-classic/da...

[fstekelenburg]

Since this a still on-going topic, and development in both worlds are going fast in two years, this might be a good moment to re-validate the status and above comparison?

I have found a rather good and lenghty article that might be of interest:
https://dzone.com/articles/opentelemetry-vs-dynatrace

[fstekelenburg]

Comparison: Dynatrace OneAgent vs OpenTelemetry Collector

1. Architecture & Approach

Dynatrace OneAgent:

- Proprietary, all-in-one agent tightly integrated with Dynatrace platform.
- Automatically discovers applications, services, dependencies, and infrastructure.
- Provides end-to-end observability with minimal manual configuration.
- Includes AI-powered Davis engine for anomaly detection and root cause analysis.

OpenTelemetry Collector:

- Open-source, vendor-neutral framework under CNCF.
- Collects telemetry data (traces, metrics, logs) and exports to multiple backends.
- Requires manual configuration of SDKs, instrumentation, and pipeline setup.
- No built-in analytics or AI—raw data only.

2. Features

Feature	Dynatrace OneAgent	OpenTelemetry Collector
Automatic Discovery	Yes	No
Dependency Mapping	Yes	No
AI/Root Cause Analysis	Yes	No
Security Monitoring	Built-in	No
Customizability	Limited	High
Standards Compliance	Proprietary	Open standards (OTLP)
Language Support	Broad, Dynatrace-specific	Very broad

3. Deployment & Configuration

Dynatrace OneAgent:

- Deployment: Single agent per host/container; minimal setup.
- Configuration: Mostly automated; advanced settings via Dynatrace UI.
- Maintenance: Managed by Dynatrace; updates are automatic.

OpenTelemetry Collector:

- Deployment: Requires installing collectors and configuring pipelines.
- Configuration: Manual setup of exporters, processors, and SDKs.
- Maintenance: Requires ongoing updates and compatibility checks.
- Scaling: Complex at large scale; needs orchestration (K8s, etc.).

4. Knowledge & Skill Requirements

Dynatrace OneAgent:

- Low barrier: Plug-and-play, minimal coding.
- Requires knowledge of Dynatrace platform for dashboards, alerting, tagging.

OpenTelemetry Collector:

- Higher barrier: Requires developers familiar with instrumentation, telemetry standards, and backend integration.
- Strong DevOps/observability expertise needed for pipeline design.

5. Financial Aspects

Dynatrace OneAgent:

- Licensing: Paid, subscription-based (cost depends on monitored hosts, metrics, and features).
- Manpower: Lower operational effort; vendor handles complexity.

OpenTelemetry Collector:

- Licensing: Free (open-source).
- Manpower: Higher effort for setup, maintenance, and troubleshooting.
- Hidden Costs: Infrastructure for collectors, storage, and backend analytics.

6. Pros & Cons Summary

Dynatrace OneAgent:

✔ Pros:
- Easy deployment and maintenance.
- Full-stack observability with AI-driven insights.
- Automatic service discovery and dependency mapping.
✖ Cons:
- Vendor lock-in.
- Higher licensing cost.

OpenTelemetry Collector:

✔ Pros:
- Free and vendor-neutral.
- Highly customizable and extensible.
- Broad language and ecosystem support.
✖ Cons:
- Complex setup and scaling.
- No built-in analytics or AI.
- Requires significant engineering effort.

7. Recommendations

- Choose Dynatrace OneAgent if you want turnkey observability, minimal setup, and advanced analytics.
- Choose OpenTelemetry Collector if you need vendor neutrality, cost control, and flexibility, and have strong in-house observability expertise.

[fstekelenburg]

Anyone to validate or comment on the above AI summarized comparison?

[Julius_Loman]

@fstekelenburg There are still many paths you can choose from. It also depends on the workloads you run. Generally, the OTEL path (using OTEL for instrumentation) involves much more work, especially when it comes to finetuning the collection. For example, when it comes to sampling of traces.

For some of the workloads, the OTEL is the path to go (if Dynatrace does not support particular technology or framework). OneAgent on the other side provides much more features than what OTEL is able to provides (bizevents capture, RUM sensors, live debugger, ... ).

I'd say - if you need any custom instrumentation (your domain specific) or metrics from your app, OTEL would be my choice (except maybe for SpringBoot/Micrometer metrics where Dynatrace did a great job with the micrometer dynatrace registry).  Some of the frameworks do have OTEL support, but there is no Dynatrace support. You should also consider your overall architecture. 

As for now, if your backend is Dynatrace, I'd choose OneAgent for all the workloads where there is a direct support. Choose OTEL for technologies where support in OA is not available. Choose OTEL if you need any custom code instrumentation or metrics to be sent. Choose OTEL instrumentation if you require immutable code (Dynatrace is added typically at runtime and instrumentation does not work without a Dynatrace server anyway). Consider OTEL cost and maintenance (mostly, you need to manage the collection layer).