YDB is a relational database. Distributed, scalable, fault-tolerant. YDB is also a platform. We offer users not only reliable transactions over tables, but also ready-made solutions based on the engine itself: persistent queues, federated queries, network disks for virtual machines (Yandex Network Block Store), and more.

YDB is used in Market, Taxi, Fintech, Alice, and other Yandex services. We already handle millions of queries per second and reliably store petabytes of data. Solutions for delivering payment data to Yandex billing and storing logs in Yandex Cloud are built on YDB.

The automation and management of distributed systems team is responsible for the operation of YDB, cloud replication, automating database deployment (cluster by button), developing an internal platform for load testing and YDB as a service. Our global goal is to make YDB as fault-tolerant and scalable as possible.

To get a closer look at the team and what we do, you can ask 10 questions to SRE!

What tasks await you

Developing fault-tolerant infrastructure for YDB and around it Our products are constantly evolving, and we are scaling along with them, so we want to launch new clusters "by button" and achieve zero downtime when migrating clusters to Kubernetes.

Along with this, we are creating a platform for load testing YDB (including for migration to AMD processors), implementing IaC (Infrastructure as Code) practices in CI/CD processes.

And one of our recent tasks is developing technical solutions for the YDB software-hardware complex, which is included in the register of PAK.

Creating an Internal Developer Platform We are improving the internal infrastructure for YDB developers to simplify their work with the database, and making our own Terraform Provider for developing YDB as a Service in Yandex Cloud and inside Yandex. We are creating our own deployment system based on the blue-green deployment strategy for the convenience of developers and minimizing risks when implementing new features.

Migrating services to Docker and Kubernetes YDB has many features that complicate the migration process. For example, we do not use a file system and work with disks directly, as block devices (therefore, we manage data writing and reading ourselves and guarantee speed and reliability). Or, for more uniform load, we use so-called client-side balancing. All of this required the development of various tools and services. Kubernetes is one of them.

We expect that you

• Can and want to write code in Go or Python (or even C++)
• Are familiar with Linux at the administrator level (e.g., Ubuntu or Debian, CentOS, Astra Linux)
• Are familiar with the TCP/IP model and have an understanding of HTTP and gRPC protocols

It will be a plus

• Have worked with K8s, Terraform, Ansible
• Are familiar with iptables at a basic level