Backend section on DUMP: Serverless, Postgres and Go, .NET Core, GraphQL and more

April 19 in Yekaterinburg will host a conference of developers DUMP . The program directors of the Backend section - Yandex Development Manager Andrei Zharinov, Naumen Contact Center Development Manager Konstantin Beklemishev and Kontura Software Engineer Denis Tarasov - told what reports await the developers at the conference.

There is an opinion that one should not expect insights from reports at the “festival” conference. It seems to us that we have made such a program from which it is worth waiting. To do this, we took only those who are deep in the subject, eliminated ⅔ applications, endlessly ruled the structure of speeches and demanded examples from the practice of the speakers.

Reports

The first two reports are related to each other, and we definitely recommend listening to both of them.
')
Problem 1. When using external API, the issue of checking incoming data is particularly relevant. One format validation is not enough; you must also make sure that the data is connected. Although the solution seems obvious, as the number of external sources grows, many individual checks easily develop into something uncontrollable. Sergey Dolganov from Evil Martians will demonstrate a structured approach to the problem, based on the use of functional programming methods.

Problem 2. To be effective when interacting with the server, it is necessary to optimize the number of calls to the API and the amount of returned data. This requires consistent entity design already at the server level. Dmitry Tsepelev (Evil Martians) will explain how this can be done effectively using GraphQL philosophy and tools, take note of the nuances and compare examples with the traditional REST.

The second block will be about Postgres and Go. Go to listen to the experience Avito and Yandex :)

You have Postgres, and you want to use Go in your project, but for the first time? This report will save you a ton of time. Software Engineer in Avito Artemy Ryabinkov will talk about the tools and all the intricacies of working with this database on Go using the example of tasks that he solves every day in Avito.

PostgreSQL and data backup? It seems that this topic has already been studied far and wide. But the knowledge will be incomplete until you know how this happens in Yandex: huge amounts of data, the need for compression, encryption, parallel processing and the most efficient utilization of multi-core CPUs. Andrei Borodin will talk about the WAL-G architecture - an open source solution for Go for continuous archiving Postgres and MySQL, which Yandex is actively developing, and you can use it in your project.

The third block is for those who are interested in speech recognition and synthesis, for whom ASR and TTS are understandable abbreviations, and for those involved in the creation of voice assistants.

Voice assistants are at their peak of popularity. Create your skill for any of them is not easy, but very simple. However, there is little known real history of this technology. Vitaly Semyachkin from JetStyle will review the capabilities and limitations of the main assistants, tell you what rakes can be expected, how they can be heroically overcome and, in general, how you can prepare the whole story. In addition, Vitaly will talk about the experience of building a “smart negotiation” based on Yandex.Station.

Advanced companies provide their APIs for building voice assistants. But what if external solutions are not available? The contour solved this problem, although the path turned out to be thorny. Viktor Kondoba and Svetlana Zavyalova will share their experience in using local speech recognition solutions in support automation, show you what you should focus on and what you can donate to increase efficiency

What else will be reports?

Recently, Redis 5 has a new data type - streams, this is the implementation of ideas from the popular message broker Kafka. Denis Kataev (Tinkoff.ru) will explain why streams are needed, how they differ from regular lines, what is the difference between Kafka and Redis streams, and also tells about the “pitfalls” that lie in wait for you.

Grigori Koshelev , Lead Software Engineer at Kontur, will consider what problems exist in the log and metrics recording if you have terabytes of data per day and also tell you about the new Open-Source solution that will make your life better.

The leader of the Kazan .Net community, Yuri Kerbitskov ( Ak Bars Digital Technologies ), will come to remind you why Application Domains is needed in the .Net Framework, and talk about what has changed when working with them in .Net Core, and how it is now live. After the talk, you will understand better how .NET Core works under the hood.

And the topic for which the most voted on the site.

A quiet revolution happened in 2014, and its echo is catching up with us. From this moment on, the infrastructure becomes completely invisible and ceases to matter. This is not about virtual machines and not about containers - they are already in the past, but about the further development of cloud service ideas - AWS Lambda (we pay only for CPU time). Using the example of a backend project in Evil Martians, Nikolay Sverchkov will tell you all about the practical side of working with serverless: how difficult it is to start, how much documentation and tutorials, whether there is support for generally accepted standards, how to test locally, how much it costs, what language is better to use, what stack tasks are most relevant.

Master Class

CTO in Mastery.pro Andrei Fefelov will hold a master class where he, together with the participants, will build a simple fault-tolerant cluster of 3 node postgres, patroni, consul, s3, walg, ansible nodes.

After the master class, you will be able to start such a cluster from scratch using the ansible playbooks provided.


All reports from last year’s conference can be viewed on the YouTube channel.

Abstracts of all reports and registration - on the conference website .

Developers, we are waiting for you on April 19 at DUMP!