Data and AI Library

Comparing the most popular open source databases:

PostgreSQL and MySQL are undoubtedly the most popular open source databases. You can check the latest developer surveys from StackOverflow and JetBrains, or DB-Engines’ popularity index if you are not convinced. Their popularity is fueled by their vibrant communities and their track record of successful deployments across IT sectors.

In this webinar, we will dive into the differences between these two popular databases and what those differences mean from a performance, operability and community engagement perspective and how Canonical can support you in adopting, securing and scaling open-source databases.

A lightweight, secure machine learning platform you can use on any infrastructure

As organisations increasingly look to take advantage of AI technologies, there are several critical considerations that they must contend with, including intellectual property, data security and costs related to computing infrastructure. Private cloud solutions are ideally suited to solving these challenges. Canonical Openstack, for instance, is a great example of a cloud platform that can be used to build and deploy machine learning applications securely and cost-effectively.

Why consider a private cloud for AI?
Private clouds are a handy solution for enterprises when it comes to AI/ML since they deliver many of the key capabilities that organisations report as important, including:
1) Cost optimisation: Private clouds enable businesses to optimise their costs by always running their workloads where it makes more sense from an economic standpoint.
2) Digital sovereignty: Private clouds offer a safe environment for data and applications by ensuring that the organisation owns access and controls the level of sharing amongst the different teams using the cloud.
3) Performance acceleration: Private clouds offer GPU virtualisation and other capabilities to improve performance and therefore project delivery, confidentiality, efficiency and time to setup as required by sophisticated AI/ML workloads.

Learn more about AI on the private cloud in this webinar:
Join the webinar on 21 February 2023, where Tytus Kurek, OpenStack Product Manager, and Andreea Munteanu, AI Product Manager, will talk more about private clouds for AI projects. The presentation will cover:
1) Key considerations when building a private cloud for AI projects
2) Performance acceleration options for private cloud
3) Guidance for Kubernetes on OpenStack for AI initiatives
And more…

Many entities mistakenly consider ML models as the only significant challenge in an AI project. In reality, one of the most essential considerations for a successful AI/ML initiative is the infrastructure on which models are deployed, tested, and rolled out. This in itself creates the need for AI infrastructure specialists, which can be both difficult and costly to find. There is, however, another option: outsourcing your AI infrastructure to a multi-cloud managed service provider.

Choosing an open-sourced managed infrastructure solution can help you accelerate your time to production, easily and cost-effectively cover some essential skill gaps in your ML journey, and even increase your budget’s predictability. In this webinar, product managers Adrian Matei and Andreea Munteanu explore the implications and benefits of adopting a Managed AI solution like Canonical’s. Join us and find out more!

Topics covered include:
- Managed AI infrastructure overview & processes
- Products & platforms
- Pre-deployment considerations
- Benefits of adopting the solution
and more!

AI is at the heart of a revolution in the technology space. From oil & gas to public sector, from telco to retail, all departments from all industries are looking for ways to put AI to work. Once organisations have finalised use case assessment, their next question is typically related to the environment they will use to develop and deploy their AI initiatives.

They often prefer the public clouds as an initial environment, because of the computing power and ability to scale as projects mature. In addition to the infrastructure, enterprises need software where they can develop and deploy the machine learning models. Open source tools such as Jupyter Notebooks or MLflow are a great starting point and enable you to have a consistent experience on any cloud for your ML workloads. When scaling projects, the need for MLOps platforms such as Charmed Kubeflow is absolutely obvious.

Join our webinar to learn more about open source tooling for AI on public cloud. Led by Aaron Whitehouse, Senior Public Cloud Enablement Director, and Andreea Munteanu, AI Product Manager, the webinar will cover:

Scenarios in which open source tooling on public cloud solves a problem for AI initiatives
Main benefits of using open source tooling on the public cloud for AI projects
Use cases from our customers
Hybrid cloud and multi cloud opportunities for AI projects

What can PostgreSQL do for AI application?

PostgreSQL is capable of efficiently storing and retrieving large amounts of data, such as training data for your models. Its wide support of the SQL standard, including SQL/MED, makes inspecting and querying that data a breeze. Moreover, PostgreSQL comes with similarity search capabilities, providing convenient index structures (GIST, GIN...) and built-in extensions (fuzzystrmatch, pg_trgm...).

Yet, there is more to PostgreSQL's ecosystem than the database server. There are more than 1000 extensions for PostgreSQL, many of which can be useful for AI developers and users, for example pgvector and PostgresML.

The Role of PostgreSQL in MLOps

MLOps projects such as Kubeflow rely on relational databases for storing different artefacts, and choosing which relational database to use for MLOps is a common dilemma. We approached a similar topic in our recent webinar “PostgreSQL vs MySQL”, where we explored how, despite looking similar on the surface, the two databases are actually very different. The significant structural differences between databases can have a tremendous influence on the performance of any ML project, so understanding the possible impact is critical for both the enterprises who use these platforms and the engineers who develop them.

So how do we use PostgreSQL for machine learning?

Join the live discussion to learn more about using PostgreSQL for AI projects. The webinar will cover:
- An overview of the PostgreSQL ecosystem for AI projects
- Benefits of PostgreSQL for AI projects
- Use cases where PostgreSQL should be used for AI projects

Learn more or contact our team: https://canonical.com/data/postgresql

Join us for a deep dive into the role of vector databases in generative AI applications. Vector databases facilitate efficient data representation, retrieval and manipulation, enabling AI systems to generate high-fidelity outputs across various domains, from natural language processing to image synthesis.

This webinar will discuss various concepts, such as generative AI, retrieval augmented generation (RAG), the importance of search engines, and efficient open source tooling that enables developers and enthusiasts to build their generative AI applications.

Learn more or contact our team: https://canonical.com/data/opensearch

As AI dominates the tech stage, the intersection of Internet of Things (IoT) and artificial intelligence is taking up more of the spotlight. From AI-powered healthcare instruments to autonomous vehicles, there are many cases where artificial intelligence can be used and benefited from on edge devices. However, building an end-to-end architecture to enable such a solution comes with different challenges to other AI stacks, both in training models and in deploying them to the edge.

## Open source at the edge
Open source solutions enable edge devices in different ways. From operating systems to ML platforms, enterprises can choose from a wide variety of solutions. This abundance of choice can be overwhelming, leading to organisations delaying decisions and not scaling their AI at the edge initiatives.
Join this webinar about AI at the edge

Join Andreea Munteanu, Product Manager for AI, Steve Barriault, VP of IoT Field Engineering and Alex Lewontin, IoT Field Engineering Manager, to discuss AI at the edge.

During the webinar, you will learn:
- The main challenges rolling out AI at the edge and how to address them
- How to secure your ML infrastructure, from the data centre to the edge
- Key considerations to start and scale your ML embedded projects
- Benefits of running AI at the edge
- Common use cases and how to get them started quickly
- Role of open source in the Edge AI space

Each type of NoSQL database offers unique strengths and is suited to specific types of applications and data requirements, making them versatile tools in modern data management. Many of them are compatible with an open source approach, giving you advantages such as the acceleration of innovation, flexibility, customisation, and community support. There are also innovations in the space where databases can run in a multi-cloud environment.
However, these features and benefits come with their own set of challenges, and operating database clusters can be difficult. In addition, databases need to be integrated within a stack of technologies, such as the infrastructure it is deployed on and other data and AI tools. Faced with these and other challenges (like unpredictable costs and industry skill gaps), companies are finding it increasingly difficult to achieve operational excellence with an in-house team to manage databases. This is why they turn to managed service providers (MSPs).

In this webinar, Canonical product managers Michelle Tabirao (NoSQL Databases) and Adrian Matei (Managed Services) will dive into what operational excellence means for NoSQL Databases, and how this can be achieved with a managed service provider. We will explore:

1) Introduction and main advantages of adopting NoSQL databases
2) Big data and AI use cases that leverage non-relational databases
3) Operational excellence in NoSQL Databases, from key operations to holistic approaches
4) Benefits and challenges when choosing a managed service provider for your NoSQL database operations
5) Future of NoSQL databases

Join us on the 18th of June 2024 for 30 minutes of expertise in NoSQL operations!

There are benefits to using a specialised database for every new use case you need to support. For example, we can make a case for using a graph database to manipulate graph data, or a specialised vector database to manage vectors. Yet, multiplying your data stores comes with consistency, cost and operability issues that you need to consider carefully.

PostgreSQL is built from the ground up to be extensible, so you can choose from over 1000 extensions to cover more use cases with the same solution – saving you time and resources.

In this webinar, we will examine the use cases you can meet with just PostgreSQL.We will cover the following:
Analytics
Online transaction processing
Geospatial data
Graph data
Search and similarity
AI and ML
Time Series and more

Learn more or contact our team: https://canonical.com/data/postgresql

Data lake is a very large scale data processing paradigm that disrupts the conventional data warehousing model. Data warehouses require all data to be structured and stored in a relational database, which can be inflexible and may require significant upfront data processing using extract-transform-load (ETL) technologies.

Data lakes can offer greater flexibility whilst retaining the benefits and efficiency of centralised data governance. With Canonical OpenStack private cloud platform, Kubernetes and Charmed Spark solutions, your data lake architecture can also benefit from extended flexibility and scalability whilst remaining cost effective to operate.

Join this webinar to learn more about the benefits of the data lake architecture, and how you can efficiently adopt this technology at scale using modern private cloud technology.

Learn more or contact our team: https://canonical.com/data/spark

AI is at the forefront of innovation. It changes the shape of all industries, challenging organisations to rethink their processes and adjust their infrastructure. Enterprises are looking nowadays to leverage their existing computing power to accelerate model training and use tools such as Kubernetes for container orchestration or Kubeflow for workload automation in order to move their ML projects to production.

In this webinar, you will:
- Learn what are the challenges with running AI on top of K8s
- Learn how to schedule your ML workloads
- Discover why K8s is a good fit for AI projects
- Leverage Kubernetes operators and schedulers to accelerate project delivery
- Understand better the cloud-native applications ecosystem and how it enhances AI/ML projects
- Have a clear path of taking ML projects to production with a fully open source solution

Learn more or contact our team: https://canonical.com/solutions/ai/infrastructure

OpenSearch is an open source search and analytics suite that developers use to build solutions for search, data observability, data ingestion, Security Event and Information Management (SIEM), vector databases, and more. It is designed for scalability, offering powerful full-text search capabilities and supporting various data types, including structured and unstructured data.
This webinar is designed to introduce participants to the basics of OpenSearch, a powerful and flexible search engine, and OpenSearch Dashboards, the associated analytics and visualisation platform.

By the end of this webinar, you will be able to:
-- Understand the basic architecture and core concepts of OpenSearch
-- Set up and configure an OpenSearch cluster
-- Install and navigate OpenSearch Dashboards
-- Create and customise dashboards to visualise data effectively
-- OpenSearch in production best practices
-- How to make it easier for you to manage OpenSearch in production

This webinar has been made for:
-- Developers who want to implement search functionality in their applications
-- Data Analysts looking to create visualisations and gain insights from large datasets
-- System Administrators responsible for managing and scaling search clusters
-- Business Intelligence Professionals interested in exploring new tools for data analysis

Learn more: https://canonical.com/data/opensearch

We’ve all heard the anecdotes of data scientists spending more time on tooling than ML models. There is no smoke without fire, right? The truth is AI/ML practitioners struggle with their environments for various reasons, including tooling fragmentation, package dependencies and access to computing power.

Data Science Stack (DSS) is an out-of-the-box solution for data scientists and machine learning engineers, published by Canonical. It is a ready-made environment for ML enthusiasts that enables them to develop and optimize models without spending time on the necessary underlying tooling. It is designed to run on any AI workstation that runs Ubuntu, maximizing the GPU’s capability and simplifying its usage.

Join us for this webinar to learn more about data science tools, with a focus on DSS and its capabilities. During the webinar, Michal Hucko, MLOps engineer at Canonical and Andreea Munteanu, AI & MLOps Product Manager, will talk about:

- Key considerations when getting started with data science
- Data science through the open source lens
- Deep dive into Data science stack (DSS)
- Demo of the DSS
- Prepare your questions and join us live to get insights into how DSS improves the developer experience for data science and ML projects on Ubuntu.

Learn more or contact our team: https://ubuntu.com/ai/data-science

Join is for a deep dive into group replication, clusterSet, and multi-cloud database architectures.

About the Webinar

Public cloud providers offer convenient services for database deployments. However, depending solely on a single provider's database-as-a-service can result in vendor lock-in and reduce your system’s resilience against widespread outages affecting the provider.
Instead, explore our approach to deploying MySQL across multiple clouds. A multi-cloud database strategy enables you to retain control of your infrastructure while enhancing overall resiliency and leveraging the unique strengths of each cloud provider.

Register to our webinar and learn more about:
MySQL Group replication
MySQL ClusterSet technology
Our reference architecture for multi-cloud MySQL deployments

Learn more or contact our team: https://canonical.com/data/mysql

Join our technical demo where Andreea Munteanu, AI/ML Product Manager and Kimonas Sotirchos, Kubeflow Software Engineer, will:

* Talk about key considerations when building machine learning pipelines
* Learn how to build ML pipelines using Kubeflow
* Give insight of some best practices when using Kubeflow pipelines

More information about the webinar:

Machine Learning Operations is a new practice that ensures machine learning projects run in an automated and reproducible manner. It is often described as DevOps for machine learning, since it brings together ML development and the operations that are required afterwards to maintain any ML initiative. MLOps has three core components: data pipelines, model pipelines and applications pipelines, which developers are often looking to optimize.

What are ML pipelines?

ML pipelines refer to the pipelines of all three components of the machine learning lifecycle and enable developers to automate ML workloads, streamlining the process of taking models to production. They are a foundational aspect to scaling machine learning projects and running them in a more effective manner.

There are multiple ways to build ML pipelines. Open source tools such as Kubeflow are an easy option, as they have already built the engine to perform such tasks. Kubeflow Pipelines are one of Kubeflow’s components which can be used to build ML pipelines.

Learn more or contact our team: https://canonical.com/solutions/ai

Ensure data security and privacy in AI applications that employ Large Language Models (LLMs).

About the Webinar
As generative AI becomes increasingly vital for enterprises – especially in applications such as chatbots utilizing Retrieval-Augmented Generation (RAG) systems – ensuring the security and confidentiality of data within these frameworks is essential.

Our upcoming genAI security webinar will address the significant challenges related to data security and privacy in AI applications that employ Large Language Models (LLMs).

During this webinar, we will introduce confidential computing as a method for safeguarding data, with a specific focus on its application within RAG systems for securing data during usage or processing. Additionally, we will outline best practices for implementing confidential computing in AI environments, ensuring that data remains protected while still enabling advanced AI capabilities.

Learn more or contact our team: https://canonical.com/data