Register

presented by Rajan Palanivel

The popularity of PostgreSQL makes it the prime candidate to be the database for a multi-tenant SaaS. But with multi-tenancy there are two fundamental issues: How do you make sure that the tenant has control over their data in a third party SaaS environment? How do you make sure that the tenant’s data is isolated from each other so that a tenant’s data doesn’t get exposed to other tenants? Letting a tenant see another tenant’s data would be considered a data breach.

Logical databases is a capability supported by Postgres for multi-tenancy. But there are infrastructure and database inefficiencies with this model, so some SaaS providers use a single database with co-mingled tenant data.

In both the models (logical databases or a co-mingled database), there is a need to cryptographically isolate each tenant’s data from other tenants. In addition to data isolation, encrypting a tenant’s data with their own keys lets tenants take control of their data, even though it resides in a third party SaaS environment.

Database-level encryption (for logical databases) or Row-level encryption (for co-mingled databases) combined with BYOK/HYOK (sometimes called Multi-tenant BYOK/HYOK) keep each tenant’s data separate and give each tenant full control over their data. Tenants can also enable/disable their keys so that they have full control on data access.

In this webinar on using PostgreSQL in a multi-tenant SaaS, you will learn how to:

• Compartmentalize each tenant’s data at the row or logical database level
• Implement Multi-tenant BYOK easily, without any application changes
• Centrally manage and define policies for keys

presented by Greg Dostatni

Effective lock management is important for any database’s performance and stability. This presentation offers an exploration of some pitfalls and solutions to the Lock Manager challenges related to large, high performance environments.

We begin by examining how table inheritance can multiply the number of locks, and how that can affect performance when dealing with more than 10,000 locks. We will present quick way of converting inheritance based partition into native partitioning scheme. The discussion then extends to native partitions and how locks can affect data archiving and partition management on a busy system. This session then explores the impact of subtransactions on locking.

Join us to learn more about PostgreSQL’s Lock Manager challenges when running on a large, busy database. We will go over some practical queries for detecting and resolving these issues.

presented by Luigi Nardi

Join us at the Materialize offices for an engaging dialogue with Luigi Nardi of DBtune about automating your PostgreSQL database.

When automating and scaling your database, there are various approaches with their own unique benefits and risks. After automation, what outcomes can you anticipate for your database?

Parameter tuning in PostgreSQL databases is necessary yet complex. Engineers frequently opt for default settings in PostgreSQL rather than risk performance issues in their databases, which are vital infrastructure components for many organizations. Consequently, they can miss out on substantial performance enhancements.

In this fireside chat, we will discuss:

• Strategies for optimizing key aspects of tuning: Observation, analysis, and implementation
• Tuning techniques and best practices for dynamic workloads
• Enhancing permanence and parameter tuning in PostgreSQL with Machine Learning
• Harnessing Machine Learning to boost your database’s performance
• The advantages of automating segments of your database