A deep dive into the TileDB data format & storage engine
For a summary and links to the Jupyter notebooks used in this presentation, see the blog post at https://tiledb.com/blog/a-deep-dive-i... .
Everything in TileDB is built around the open-source array storage engine, TileDB Embedded. Any data can be universally modeled as multi-dimensional TileDB arrays, and it all starts here.
Complete with code examples and performance tips, this presentation provides a thorough technical primer for getting started with TileDB Embedded. It also lays the foundation for getting started with the universal database, TileDB Cloud. You will learn:
• Why we chose arrays as the foundational data structure of TileDB
• How dataframes are just a special instance of arrays
• Data layout, tiling, compression and other filters
• Dense and sparse array indexing for rapid slicing
• Data versioning and time traveling
• Why TileDB is “cloud-optimized”
• How to achieve unprecedented performance
• Going beyond storage: compute push-down
• How to evolve and manage schemas in TileDB
Contents of this video
0:00:00 – Introduction & Why arrays
0:13:15 – The basics
0:32:46 – Tiling & layout
0:42:01 – Advanced internal mechanics
0:59:18 – Performance tips
1:02:43 – Work in progress
1:04:37 – TileDB vs. others
1:08:06 – Q&A
About
TileDB makes data management and compute fast, easy and universal. Manage any data as multi-dimensional arrays and access with any tool at global scale.
Connect with us
Website: https://tiledb.com/
Twitter: / tiledb
LinkedIn: / tiledb-inc
Book a personalized product demo: https://tiledb.com/demo
Sign up at https://cloud.tiledb.com/auth/signup and contact [email protected] for free credits.