In this blog post on Percona Live 2018 sessions, we’ll talk with Jun Su, Principal Engineering Manager at Microsoft about how Microsoft built MySQL, PostgreSQL and MariaDB for the cloud.
Offering MySQL, PostgreSQL and MariaDB database services in the cloud is different than doing so on-premise. Latency, connection redirection, optimal performance configuration are just a few challenges. In this session, Jun Su walked us through Microsoft’s journey to not only offer these popular OSS RDBMS in Microsoft Azure, but how they are implemented in Azure as a true DBaaS. We learned about Microsoft’s Azure Database Services platform architecture, and how these services are built to scale.
In Azure, database engine instances are services managed by the Azure Service Fabric, which is a platform for reliable, hyperscale, microservice-based applications. So each database engine gets treated as a microservice. When coupled with Azure’s clustering — a set of machines that the Service Fabric stitches together — you can scale up to 1000+ machines. This provides some pretty impressive scaling opportunities. Jun also walked through some of the issues with multi-tenancy, and how different levels of multi-tenancy have different trade-offs in cost, capacity and density.
After the talk, I spoke briefly with Jun about Microsoft’s efforts to provide the different open source databases on the Azure platform.
