In mid-June, a conference was held in our office
For 9 hours in two halls of T + Conf 2019, it was possible to listen to 16 reports. We talked about how Tarantool will develop further, how this DBMS can be used in a harsh enterprise. There were many practical Tarantool reports: about the protocol for building a cluster, about providing omnichannel, about caches and replication, about scaling. And about a third of the speeches were about practical examples of using Tarantool in different companies and for solving a variety of problems.
For example:
CI/CD applications on Tarantool: from an empty repository to production
Konstantin Nazarov
Konstantin spoke about a new approach to structuring and delivering applications in Tarantool:
- how to manage dependencies (rockspec + friends);
- how to write and run unit and integration tests;
- show a preview of a new test framework for applications;
- how to package applications along with dependencies (and why we chose static linking);
- how to deploy to production with systemd.
Presentation
Tarantool: now with SQL
Kirill Yukhin
The report is devoted to Tarantool architecture and its evolution. Kirill told why it is important to have a database and an application server in the same address space, why Tarantool was made single-threaded, and why a database-in-memory mechanism for storing data on disk is needed. Then Kirill spoke about the latest developments of the Tarantool creator team: why we added the SQL syntax and how it can solve your problems.
Presentation
Benefits of Tarantool Enterprise
Yaroslav Dynnikov
Tarantool Enterprise is not only a valuable mech, but also a feature-rich SDK. Yaroslav told how NT differs from the open source version and what benefits it can bring. And there are a lot of differences in it: this is a cluster administration toolkit, and a ready-made development workflow, and a static build that is not demanding on setting up the environment.
Presentation
Tarantool vertical scaling using Intel Optane
Georgy Kirichenko
George explained how to use Intel Optane with Tarantool. Considered the effects of using Non-Volatile mode for writing transaction logs, the possibility of vertical scaling of the In-Memory engine in conjunction with Intel Optane Volatile mode, good and bad load profiles in terms of throughput and latency. And George will also talk about different implementations of Intel Optane and compare them in relation to Tarantool.
Presentation
SWIM - cluster building protocol
Vladislav Shpilevoy
SWIM is a protocol for discovering and monitoring cluster nodes, propagating events and data between them. The protocol is notable for its lightness, decentralization and independence of operation speed from the size of the cluster. Vladislav talked about how the SWIM protocol works, how and with what extensions it is implemented in Tarantool.
Presentation
In general, there was a lot of useful information!
If you could not come to T+ Conf 2019, or want to refresh some moments in your memory, then
All our photos from the conference (maybe you will find yourself on them):
We donβt say goodbye on this, but we are waiting for you next year at T + Conf 2020, stay tuned for announcements!
Source: habr.com