OTEL Me Why: Why am I very excited about otel

News Fetcher15 hours ago

0 0 7 minutes read

Limited time view! I am ready for my next adventure as Devrel/Evangelist/Talespinner. If this looks like something you need, then drop a line for me By e -mail Or on LinkedIn.

My blog about pricing from the other day drew the attention of people in Metricfire, and we conversated some ideas, ideals and challenges that revolve around monitoring, observation and its wider landscape.

At some point, JJ, the main engineer, asked, “I have been blogging about preparing for a certificate of measurement from the open distance. What is going around Ollete that I was very excited?”

I gave a quick answer, but JJ’s question made me think, and I wanted to put some of these ideas here.

Otel is the best thing since …

Allow me to start answering the JJ question directly: I find an exciting open dimension because it is the biggest change in the way in which monitoring and observation have been made since the effects (which appeared around 2000, but it has not been widely used until 2010-ais).

The effects were the biggest change since … at all. Let me explain.

See this picture? This was what was the case using monitoring to understand your environment when you started nearly 30 years ago. What we wanted is to know what was happening in this boat. But this was never an option.

We can get rid of the scales together from the networks of the network and OS and we can build some DB software that gave me more insight. We can collect and collect registry messages (with a lot of work) together to discover trends across multiple systems. All this would give us an idea of how to run the infrastructure and conclude things that may occur alongside. But we really didn’t know.

Track all of this. Suddenly, we can get solid data (and get it in actual time) about what users were doing, and what was happening in the application when they did so.

It was a complete change in the sea (intended from the pun) of how our work and what we watched. However, tracking did not remove the need for scales and records. And famous (or notorious) “three columns” of observation.

Recently, I started working through the book “Learning OpenTrelmetry”, and one of the comments that surprised me is not “three columns”, meaning that it is not united to formulate a uniform. The authors TED Young and Austin Parker re -framing a mixture of scales, records and monuments as “the three browser tabs” because many tools put the voltage again on the user for the face between the screens and put them together by sight.

On the other hand, OTEL outputs can offer all three data flows as one “braid”.

3 lines bearing a tracking mark, standards, and records with interrelated lines to form a braid.

It should be noted that despite OTEL’s ability to combine and linking this information, the authors of the book later indicate that many tools still lack the ability to provide it in this way.

Although it is a continuous work (but what, in his world, not?

Otel is Esperanto From monitoring

Almost each seller will jump a chance to send all your data to them. They insist that they have one real note tool.

A single ring image of the head of the rings, with the words over "One of the tools for all of them, one tool to put it in a pipeline, and one tool to scrape all of them, and in the sales seller connects them.

In fact, let’s go out in the open: there is no simply unique “better” monitoring tool there is more than there is more than one “better” “” “” better “programming language, car model, or pizza style.* There is no single tool that covers 100 % of your needs in each use.

For the largest tools, the use cases that are not part of its absolute sweet spot will cost you (in terms of hours or dollars) to correct.

Therefore, you will have multiple tools. Needless to say (or at least) you will not charge a full version of all your data to multiple sellers. Therefore, a large part of your business as a monitoring engineer (or a team of engineers) is to set the remote measurement for the cases of use they support, and thus to the tools that you need to use in these cases of use.

This is not actually a difficult problem. Certainly, it is complicated, but once the appointment is obtained, achieving this is relatively easy. However, as I like to say, the cost of buying the puppy is not the problem, it is the cost of continuing to feed it.

Because the tools that you have today will change on the road. This is when things become crazy hard. You have to hope that things are well documented enough to understand all remote measurements for use.

(Narrator: They will not, in fact, are well documented enough)

Then you also have to hope that your devices be documented and understood well enough to see how to cancel the X tool and tools to keep the same capabilities.

(Narrator: This is not how it will decrease.)

But OTEL solves both “purchase of the puppy” and “feeding the puppy”. My friend Matt McDonald Wallace (Solution Engineer in Gravana) and put him in this way:

Otel does not solve many problems about it “O great! Now we are trapped with the X seller and millions cost us to reshape all this symbol” instead of “Oh, we change the sellers? Wonderful, allow me to only update my ending point …

Not only that, but OTELS’s ability to create pipelines (for those who do not reach this concept, it is the ability to determine, filter, sample and convert a stream of data before sending it to a specific destination) means that you can send the data flow itself to multiple sites selectively. This means that your security team can get an unparalleled raw system while still burning. Some data can go – effects, records and/or standards – to one or more seller.

That is why I say: OTEL is Espierto of observation.

OTEL sauce is not otlp

… it’s unification.

Before I explain why the real benefit of OTEL is otlp, I must take a second to explain what otlp:

If you search for “What is the remote measuring line protocol?” Maybe you will find some differences “… a set of standards, rules and/or agreements that define how to send otel data from the thing that you created to a destination.. This is technically true, but not very useful.

Functional, OTLP is the magic square that takes standards, records or effects and sends them where they need to go. It is not a low level, for example, TCP, but in terms of how to change the monitoring engineer day, it may be. We are not Use Otlp as much as we are Indicate It should be used.

Just to be clear, OTLP is amazing and amazing. Not only (in my opinion) is no less important than some other aspects.

No, there (at least) two things, in my opinion, made otel such an evolutionary shift in monitoring:

University

First, it unifies the model of the presence of a three -level mosque (not an agent) in the middle, brown. For us the elderly in the surveillance space, the idea of a mosque is not new. In an era in everything on everything, she was unable to stay away from a thousand (or even a hundred) agents talking to some distant destination. The shift to the structure of the cloud has changed all of this, but it is still not the best idea.

The presence of one number (or a small number) of balanced systems in the download that takes all data from multiple goals-with an additional benefit is the ability to process these data, filtering, samples, or plural, etc.-before sending them forward, is not just a good idea, but can have a significant direct impact on this result.

Connotations

Look, I will be the first to tell you that I am not the best developer in the world. Therefore, the issue of semantic terms does not usually keep me at night. What Do Keep me awake is the inability to get part of the data that I know should be there, but not.

What I mean is that it is somewhat common that the data point itself – for example the frequency range – is referred to as a completely different location and location on the devices of different sellers. Perhaps this does not seem very strange.

But what about the same data point, which differs on two different types of devices from the same seller? It is still not strange?

Let’s talk about the same data point that differs in the type of device from the same seller, but two different models? Get strange, true (not to mention annoying).

But the real Kicker is when the same data point varies in two different parts of the same device.

Once you descend this rabbit in particular, you have a completely different estimate of the semantic name. If you are looking for a central processing unit, frequency range, cumin, or anything else, I really like to call it the same and is found on the same location.

Otel does this, and does it as a basic aspect of the platform. I am not the only one I noticed.

Several years ago, during a meeting between the supervisors of Prometheus and the Opentil measurement, he mocked the incentive of Prometheus, whose name was not revealed. It might seem a little ridiculous, but it is also true.