DocumentDB Extension: An Open Document Database on PostgreSQL

MongoDB has become the standard API for document databases, and today’s users expect more freedom, open source solutions, and portability. Over the years, the ecosystem has responded with various alternatives, including Amazon DocumentDB, which emulates MongoDB on AWS, JSONB and the SQL/JSON standard in PostgreSQL core, libraries like Pongo, and compatibility layers such as FerretDB.

Recently, Microsoft open-sourced DocumentDB, a MongoDB-compatible extension for PostgreSQL. This solution, proven at scale in Azure, is now governed by the Linux Foundation to build an open, vendor-neutral community and rely on the PostgreSQL ecosystem.

This presentation goes through how this extension actually works. It adds BSON data types and extended RUM indexing to PostgreSQL, along with a document API, and a Rust-based gateway implementing MongoDB’s wire protocol. This is a key difference from JSONB-based approaches, as compatibility is handled at the protocol level, not only through SQL. We will also look at how indexing evolved, what changed recently, and the trade-offs involved—when a document model makes sense, and when it doesn’t.

Linux load average and other silly metrics

Databases are predominantly run on Linux operating systems in cloud and on-premises environments. Metrics such as %CPU, load average, I/O wait, and free memory are commonly used to measure performance. However, interpreting these metrics correctly can be challenging. In this interactive live demo, we will run a workload and explore the intricacies of performance monitoring using the "top" command. We will also discuss the limitations and common misinterpretations when relying solely on these metrics.

Hibernate: Mapping Strategies and Their Database Performance Impact

Your database administrator has raised concerns about inefficient queries generated by Hibernate without providing any explanation? This can be challenging because ORMs abstract away the complexities of what is executed in the database, and every database has a unique implementation. The communication gap between developers and operations further complicates the issue.

In this session, we will explore some Hibernate mapping strategies and analyze the SQL queries generated by them. We will also discuss the impact of these strategies on database performance and provide insights for Java developers to optimize code execution or improve their indexing strategy. Developers can improve application efficiency by understanding how their queries work on the database.

YugabyteDB: Distributed PostgreSQL on Kubernetes

Unlike traditional SQL databases, with their monolithic architecture, YugabyteDB brings horizontal scalability and resilience to the table. Our demo on Amazon EKS will delve into the core motivations behind adopting this cloud-native DB: elasticity and resilience, all while remaining PostgreSQL compatible and open-source.

Anatomy of a distributed SQL database (YugabyteDB)

Porting all the features of PostgreSQL to a distributed database that “scales” horizontally is a challenge. But also the opportunity to modernize the underlying technologies of the DB, which becomes “cloud-native”: consensus protocols, logical clocks, automatic sharding. And to replace B-Tree indexes with LSM Tree and SSTables, more suited to SSD and distributed storage.

YugabyteDB is open-source, and we will go into the details of the architecture, at the crossroads of PostgreSQL, Spanner, Cassandra, RocksDB… to better understand the reasons for a new database, and its underlying technology. We will discuss the advantages and the challenges of this unique architecture design: re-using the PostgreSQL query layer, plugged on top of a distributed storage and transaction layer

Document Data Modeling and Denormalization in SQL and NoSQL

As database technologies evolve, the distinctions between SQL and NoSQL become less clear. NoSQL databases like MongoDB may recommend some referencing and joins rather than embedding all entities into one document. At the same time, SQL databases allow denormalization and can store and index JSON documents. Some even provide MongoDB-compatible APIs for querying relational data, as seen in Oracle Database and PostgreSQL when used with FerretDB.

This session explores the core principles of data modeling. Should you embed multiple entities within a document or define references and foreign keys? What factors influence this decision: performance, flexibility, or query access patterns? We will examine the trade-offs between normalization and denormalization using examples relevant to SQL tables or NoSQL collections. Whether you are working with documents or rows, this talk will give you a deeper understanding of the appropriate modeling approach to persist your application objects.

Beyond PostgreSQL: Distributed SQL with YugabyteDB

Postgres is a widely used open-source database for OLTP, and its powerful SQL features are continually improving. However, it faces operational challenges such as process per connection, vacuum issues, resilience to failure, and downtime for upgrades. These issues require a different storage architecture. Distributed SQL, inspired by Spanner, addresses some of these issues but often has limited SQL support. It is also essential to remain Open Source when offering an alternative to PostgreSQL

YugabyteDB overcomes these limitations by utilizing PostgreSQL code for SQL processing on multiple active nodes and employing distributed transactional storage to scale horizontally. This marks the evolution of databases towards Distributed SQL databases. We will demonstrate YugabyteDB's resilience and scalability.

This session brings the concepts of Distributed SQL and a live demo showing PostgreSQL compatibility, elasticity, and resilience

MongoDB API: Getting Developers to Love the Database Again

MongoDB has become popular among developers due to its ease of use compared to SQL. Many SQL databases have also introduced a MongoDB API layer, including Oracle Database with OSON, FerretDB for PostgreSQL, MariaDB with MaxScale, and SAP HANA.
In this session, we will demonstrate how using a MongoDB API enhances the developer experience for building modern applications compared to using SQL. We'll also cover fundamental concepts to help us better understand the developer's perspective.
Additionally, we will explore some of the deeper internals, comparing how MongoDB functionality is emulated in SQL tables within Oracle Database versus how it is executed natively in MongoDB with WiredTiger.