author profile pic

By Eleanor Bennett

Resources

3 min read

This article will give you a quick overview of some of the key attributes you should know in order to get started with leveraging the OpenTelemetry collector for your next telemetry project.

As an integral component of any project that involves distributed tracking, the OpenTelemetry Collector plays an important role. Simply put, it is helpful to know that the collector itself is a data pipeline service that collects telemetry data.

It is true that the functionality of the OpenTelemetry Collector is entirely optional in many cases. However, it can be extremely valuable when used for collecting information about your infrastructure. Furthermore, in addition to being able to ingest signals in a variety of formats, the OpenTelemetry Collector is also capable of sending the data to a variety of observability backends as well as transforming the data within the same process.

Essentially, the functionality of the OpenTelemetry collector can be broken down into three distinct components. As part of the OpenTelemetry collector, there are three main components, namely processors, exporters, and extensions. Using the OpenTelemetry collector, you can receive, process, handle, expose and export telemetry data in a completely vendor-agnostic manner.

Telemetry data (including metrics, logs, events and traces) can be received from a variety of sources. It is possible to transform and process the collected data in a variety of ways once it has been collected. In order to receive data, multiple formats and tools can be used, including Jaeger and Zipkin, OpenCensus, and even the dedicated OpenTelemetry protocol. Upon receiving data, the data can then be consolidated into a single pipeline that will then be used to process all the data. It is at this point that we will move on to the next component of our analysis, the processing.

In the data pipeline, there are many attributes where the use of processing can be beneficial in order to highlight key metrics. During this stage, you will be able to use processing to refine, augment, and transform the data in order to meet your particular observability requirements. As a general rule of thumb, a processor pre-processes the data before it is exported, and it can also ensure that the data is successfully loaded into a pipeline by ensuring that data goes through the pipeline successfully.

Different tasks may be performed by a processor at different stages of the process. For instance, this component may conduct sampling, update various attributes or modify some of the attributes based on the data points that it is collecting.

In the final step of the process, exporters are used to export the telemetry data to the desired backends. Depending on the systems and vendors that you are looking for, there are a lot of relevant exporters that are available out there for you to choose from. Most of the commercial proprietary vendors out there have exporters for pretty much the vast majority of their series. It is possible to export data from the Opentelemetry collector to a final destination (for example, Jaeger or Prometheus) during this final stage.

A common difficulty users run into when working with Jaeger, is that they often find that there is a very limited number of ways in which they are able to find what they're looking for when working with this software. Depending on the situation, you may wish to use a solution such as the one provided by OpenTelemetry for Logit.io in this case.

How Many Events Per Second Can The OpenTelemetry Collector Handle?

Based on the expert opinion of our team, a single collector working with a small amount of CPU memory would likely be able to handle thousands of events every second with no problems at all. The reason for this is that the components of the collector are fairly robust in terms of design.

What Is The Most Efficient Protocol To Send Data To The OpenTelemetry Collector?

You would expect to find a range of solutions recommended for different programming languages, however, the general rule of thumb is to use gRPC, regardless of the programming language used.

In addition to allowing users to collect trace data automatically, the gRPC Instrumentation enables users to export this data to a backend of their choice. Ruby is the only programming language that does not support the gRPC protocol, with most other languages supporting it.

If you enjoyed this article why not read our guide to Prometheus monitoring tools or Opensearch vs Elasticsearch next?

Get the latest elastic Stack & logging resources when you subscribe

backReturn to Blog

© 2022 Logit.io Ltd, All rights reserved.