Selection of best articles about Tarantool

The Tarantool can process up to million requests in a second. Other database services can process only tens of thousands requests at best. To some people who don't know the subtleties this seems to be suspicious. But the speed of the Tarantool has real technical reasons and bases on combination of algorithms. There are no philosophical disputes or holivars on topic "to each his own database" you can find. But the article will guide you on truth why is Tarantool so fast in starting, reading and saving your data, how it circumvent pitfalls of snapshotting and system calls.

Inexperienced managers dislike commercial options including Tarantool. Often they look for cheaper alternative, but as a result they get expensive, slow and awkward system instead of saving money. But in fact Tarantool costs quite less than its analogs. This platform helps to business save millions dollars. How is it possible? Read in this case. You will learn how to calculate the project budget and which option to choose. Also this article will help you with pitfalls in cold and hot data usage.

Tarantool is most associated with high loaded systems. And what's going to happen with simple app with data recording and processing if we will create it on Tarantool and load it the most? Are there reasons to start simple apps on ultra fast Tarantool, you can find out in Vasily Ozerov's experiment. This article will help you repeat this test step by step and familiarize with Tarantool from installation to app launching and speed testing.

You want to try Tarantool but hesitant? This article will help to overcome dithering. You will see step by step how developers started high loaded app on Tarantool version 2.2 from scratch. Namely, they created net queue with ability of delayed processing, task auto-return by triggers and statistic sending to Graphite via TCP. As a result, there are about 300 code lines in this app, written in one day. This queue can stand transmission of more than 20 thousand messages per second. Save this article in bookmarks and return when you need to prepare the environment, write the queue or add the pending processing.

The more project is successful, the more it has data. The more it has data, the more the load to the server. When left alone, it runs too long. The logical solution is to scale. There are convenient framework for scaling to make some copies on different servers or to split data to different nodes of Tarantool. This framework allows you to create and configure clusters of multiply Tarantool instances. This manual will help you to manage Tarantool in Kubernetes cluster. There are step-by-step algorithm from development to operation.

Article will coach you to deploy Tarantool-based distributed applications and manage them. There are training by the real case. Tarantool allows fast update apps inside cluster, auto-switching data recording to replica if leading instance process goes wrong. Built-in authorisation module solves major security issues without complex setting. Application configuration is determinate by yml-file can be easily managed for the issue. Read to know how to set up Tarantool cluster topology.

Monolithic systems of demanded projects rest against the top of capacity and stall business processes. That's how it happened at the Wildberries, when the company launched Black Fridays so the workload tripled. To distribute the load of monolithic functionality, the company divided the system into microservices. It wasn't all in easy way. Read in the article technical details of how the company switched from a monolith to microservices and conquered all difficulties.

Major innovative projects suffered from lack of abilities of ready-made systems. The trending technology of the industrial internet of things without failures can be attained only by scalable and physically independent platform. Mail.Ru IoT Platform manages to create such the platform from scratch. For now this platform allows to collect data from thousand devices at the same time and process thread in near real-time. Among others, it uses custom rules written in Python and Lua. But it wasn't easy. In this article you can read about challenges to all stages from raw idea to operations.

It is impossible to choose a database management system suitable for a business project without understanding how they work. It is not necessary to go through a detailed course - you will make the right choice if you delve into this article. Here you will learn the principles of architecture at the macro level and the specifics of engineering, the algorithm and data structure of Tarantool. In the first part, you will understand how to solve macro problems, and thereby create space for yourself to choose at the middle level and micro level. And then, using specific examples, you will see how to implement the fastest possible data structures.

In release 2.6, the Tarantool in-memory computing platform has acquired synchronous replication. Before that, Tarantool had built-in asynchronous replication as a horizontal scaling tool, and it was enough for most tasks. Why synchronous replication was added, what it gives users, you will learn in the article from the Tarantool developer. Here you will also find an algorithm for implementing synchronous replication for the life cycle of a transaction, and understand the differences in the Raft algorithm. You will also get acquainted with the interface for working with synchronous replication in Tarantool.