At each major regular conference there are speakers who come every year, tell something new and always like the audience. It is very difficult (and why) to always be in the very top for a person who does not perform professionally, but always delivering confidently good material is real. One of the speakers who made several successful reports on Highload ++ and RIT ++ is Ivan Kruglov from Booking.com.

A few days ago on the Ontiko blog there was already an article on the preparation of speakers, devoted more to the presentation of the material, and today I would like to talk about another aspect of the preparation, which I mainly do at RIT ++ and Highload ++. Let's look at the example of Ivan’s last speech, what’s important and what we are working on with the speakers in the preparation of the conference.



Slides here .
Of course, any report about microservices, or at least about SOA, automatically gets +2 to charisma, but the speaker should still be able to take advantage of this.

Disclaimer: About SOA is only a parsed report, not the article itself.

Plot

Formulation of the problem

Is it true that among us there are few lucky people who are now developing their project from scratch, and already at the design stage they laid the microservice architecture in it? Ok, let a lot. Ok, even if not all of these people are lucky, because some have definitely laid the microservice architecture in vain and will receive all the inconveniences associated with it right at the start, and their project will never grow to the problems that it solves .

We’ll not think about these people, whether they are lucky or not, but about others, who are also many. About those who inherited something big, monolithic, old and very fragile, and who try to cut it into microservices not at the behest of fashion, but out of hopelessness.

It seems to me that when all these people hear “adish, monolith, PHP”, when they see an attempt to color the whole architecture in two technology stacks (“old” and “new”), and this attempt is immediately recognized as unsuccessful - they actually receive a signal “You and I are of the same blood”, or, less pathetic, “I am the same person as you, and I have the same difficulties.” This is an important signal, it gives the speaker an additional credit of trust from the audience, and they immediately understand what kind of decisions to expect from the report.

Of the possible improvements, I note the following: the problems that Ivan refers to at the beginning of his speech are very common. For better focus it was worth narrowing them. For example, the section on traffic management difficulties (minute starting at 40:42), in my opinion, asks for a statement of the problem, not a conclusion. After all, it is about how to manage traffic with the help of envoy and a self-written control plane, which is discussed in the report.

conclusions

It is useful to repeat the main thoughts that we want to convey to the audience at the end, as in this case the speaker does. This is necessary in case people during the story for some reason could not isolate the most important. Again, there are quite a lot of this important, it is useful to list again.

In the conclusions of Ivan in this speech, I am confused by the fact that only three and a half follow from the three points directly from the story. See (start at 40:00 ):

1. Pay special attention to the interaction between the services, and all the bazvords that are invented around this are not just bazvords, but really necessary things. This is true, but it follows, rather, from the previous report (in RIT ++, here is the record ) than from this.
2. The pros and cons of approaches with a library or with a local proxy in the conclusions are presented in approximately the same detail as in the presentation itself, and in approximately the same vein. That is, this conclusion rather repeats what was said before than summarizes it, therefore we will set it off for half.
3. Envoy and Istio are useful, interesting tools with which you can establish interaction between services. Yes, the main part of the report is about this.

For those who looked at Ivan’s previous reports, the conclusions should go fine, but those who watched only this one could feel a certain gap.

Slides

Schemes

I never get tired of reminding that the audience has a battery in their head, which is spent on understanding the speaker. This is similar to the principle of saving fuel: unproductive battery consumption should be reduced. Not all the power of the mind of the viewer is directed to the report. Someone has a child, another may have a tooth, someone has an appointment for the evening, someone else has fallen, and he just received an SMS about it. People’s attention is divided between these important things for them and our performance. I want them to at least not have to spend energy deciphering what we are trying to say.

Complex schemes are one of the most energy-intensive types of content. Look at slide 36: really, it takes a lot of strength to figure out what's going on here?



Now look, if not yet, how this diagram appears in the presentation. The story about her begins at 24:20 and lasts almost 4 minutes. At the same time, it is always clear which element of the circuit Ivan is talking about at the moment, since the elements appear sequentially. The second bonus of successive appearance is that there is a dynamic. When some events regularly occur on the screen, this is better than a static picture, as each new movement attracts the viewer's eyes.

Any scheme (ok, almost any) has an entry point, starting from which this scheme needs to be considered and told, and further the narrative according to the scheme somehow spreads. It is in this order that the schemes in Ivan's story open, and when preparing reports, we recommend that everyone do this.

Slides as an addition to the speaker

There is a good joke with a fair amount of truth: how to understand if you have good slides or not? If you can figure out what you are talking about without them, then the slides are definitely bad. Many speakers strive to make self-contained slides so that they can be read, and this seems like a good idea until it turns out that it is difficult to speak with such slides: they fight with the speaker for the audience’s attention.

In this presentation, everything is fine with this: the slides are clearly designed to illustrate the speaker’s speech. For example, there would always be headlines in “traditional” slides, and they only take place in illustration slides. The third one has the title (and other attributes typical of the “traditional” slides), and the next slide, which has them, is the twenty-seventh.

That's right: there is a story, the main element of which is the speaker, and there are illustrations for it, and not two competing versions of one story. And if we want people to read something, let's write an article on Habr.

Projector and contrast

It is better to check the slides in the hall where you are going to speak in advance. Unfortunately, this is not always possible, and at the Skolkovo business school, where Highload ++ and RIT ++ are held, the picture in different rooms is different. If we are using a projector, then in the general case it is better to assume that it does not give high contrast.

As a horror story, look at this photo taken just during the report. No slides are visible on the left side behind the speaker . Of course, in reality, the screen looks a little brighter than in the photo, but you can understand the essence of the problem:



Screenshots from monitoring, which for some reason are all colored on a black background, are especially affected on the projector (I’m sure that most of the monitoring has a bright gamma in the settings, but apparently the evil system administrators do not use it). That is, slides like number 44 have a particularly great risk of turning into a pumpkin:



Here we are faced with a philosophical question, which is more convincing: a picture from the instrument’s tool, which is hard to see, or an obviously home-made graph supposedly from the same data. But if you cannot replace the picture for some reason, then it makes sense to carry out the work that has been done on the slide, namely: select only an interesting place, stretch the schedule as much as possible, get rid of the legend and, if it is important, manually make larger signatures to axes.

About rehearsals

I will criticize Ivan for the fact that he often interrupts himself to give a previously forgotten piece of information, or stops to pick up the wording, and this process often happens out loud. These gaps, it seems, are not so many as to interfere with monitoring the course of the story, but enough to make them noticeable and slightly annoying.

This time at Highload ++ 2017, the preparation activities were not quite the same as usual (experiment), and there was no complete run of this report with a record that could be viewed, and this would greatly help to make the story’s logic more continuous .

Some experts do not recommend watching a recording of their rehearsal on the eve of the performance, believing that this does more harm than good. Harm can happen if, during viewing, you pay the main attention to the fact that it is still impossible to fix it in a day. I have repeatedly seen the speaker, when he first saw himself in the recording, almost wring his hands with a groan: “It was terrible ... Solid uh, there and how, what to do? ..” If you think about it live in front of the audience, it really will only get worse.

You can’t get rid of uh-ee in one day (more precisely, I don’t know how to do this). The secret here is that in our speech we notice much more parasitic words than in someone else's. And in the opposite direction, this also works: it seems to us that everything is terrible, and an outside observer perceives us quite normally. Speech clogged so that it becomes a real problem is quite rare.

Therefore, when viewing, you need not think about the white monkey, but look for holes in the logic. If you expressed it incomprehensibly or ambiguously or clearly could not immediately formulate a thought, you need to stop the recording and think over the wording. Of course, not everything that is thought up will be remembered before the real performance, but it will be much easier. Therefore, when preparing conferences, we give speakers, among other things, the opportunity to speak on camera, get feedback and then analyze your text. From this point of view, it is important to watch the recording, since it is impossible to remember all the places in which difficulties arose directly during the rehearsal.

Come to participate

In general, if you want not only to speak, but also to pump your skills in the preparation process, then it is definitely worth applying for a report to the RIT ++ conference festival .