July 2022 Summaries

Preset has announced its partnership with Databricks, making Preset Cloud available through Databricks Partner Connect, allowing users to easily launch a free instance of Preset Cloud with pre-configured database connection details directly from their Databricks account. This integration significantly streamlines the process of connecting Databricks to Preset, reducing the steps from 20 to just 6. Both platforms, built on open-source projects, enable a unified data environment where users can employ Preset’s no-code chart building for interactive dashboards without SQL knowledge, while Databricks offers a Lakehouse platform that merges data lakes and warehouses for enhanced reliability and flexibility. Preset extends Apache Superset with features like multi-workspace management, single sign-on, and a powerful SDK, advocating for open-source solutions in business intelligence.

Using Google Sheets and Preset Cloud, users can create data visualizations and dashboards without needing access to a traditional database. By uploading spreadsheets to Google Sheets, users can treat them as databases within Preset, following guidelines to ensure data is tidy and properly formatted. The tutorial outlines steps for connecting both private and public Google Sheets to Preset, using a public NBA games dataset as an example. Users can create various chart types, such as time series bar charts and graph charts, by selecting appropriate metrics and columns. Preset uses the SQLite dialect of SQL for querying Google Sheets data, allowing users to write custom queries in the SQL IDE. The tutorial also covers dashboard customization, including resizing charts and adding dynamic filters, and encourages users to explore Preset’s wide range of chart types and collaboration features.

Superset 2.0 marks a significant advancement for the open-source project, enhancing user experience and stability by removing outdated feature flags and code paths, and introducing new features such as a horizontal bar chart and an improved visualization picker in Explore. The release also brings enhancements to database experiences for platforms like Databricks, Trino, and Pinot, and introduces a native database driver for Databricks. Users can enjoy a no-code chart building experience with quick visualization switching, a drag-and-drop feature for table charts, and mixed charts supporting advanced analytics. Additionally, quality-of-life improvements have been made, such as a redesigned Run and Save button and a more consistent results panel, while numerous bugs have been addressed to ensure a smoother user experience. The release includes new Jinja macros for enhanced querying capabilities and offers multiple upgrade options through PyPi, GitHub, and the Apache Software Foundation, with additional resources like release notes and tutorials available for users.

Preset Cloud is a business intelligence platform that enables teams to create interactive dashboards for quick and easy data exploration, while also supporting custom data visualization applications through Superset as a data access layer. Developers can leverage Preset to manage database connectivity, authorization, and caching, allowing them to focus on presenting data effectively. The platform provides various tools, including APIs, a CLI tool, and a Python SDK, to facilitate interactive and programmatic querying of databases and datasets. With features like the Preset CLI for executing SQL in REPL mode and Python scripts for complex data manipulations, users can seamlessly integrate Superset's capabilities into their workflows. This approach allows developers to build tailored data visualization solutions by embedding Superset components into applications, thereby enhancing their ability to create dynamic and user-responsive data experiences.

Slow dashboards are a common issue in business intelligence tools, often resulting from the complexity of the underlying SQL queries generated by platforms like Apache Superset or Preset. To enhance dashboard performance, it is crucial to focus on optimizing databases for querying, as this has the most significant impact compared to other factors like enhancing dashboards for viewing or improving access. Key practices for database optimization include using materialized views or tables to pre-calculate and store aggregated data, partitioning large datasets to facilitate quicker access to relevant data, and implementing indexing to speed up data retrieval. Additionally, filtering within the database can significantly reduce the data scanned, thus improving query execution time. The choice between OLAP and OLTP databases can also affect performance, with OLAP being more suitable for complex, multi-dimensional data analysis and OLTP for handling numerous transactions efficiently. Overall, while each database system offers unique optimization features, aligning pre-calculated datasets with user visualization needs can substantially decrease query run times.