When You Can't Avoid a Seq Scan: Data Skipping in the New Tarantool Columnar Engine

on YouTube ↗48m

" summary

The talk explains data skipping as an alternative to full table (Seq) scans when precise indexes aren’t available, focusing on its implementation in Tarantool’s new columnar engine. It compares index efficiency in the columnar engine versus the row store and describes how data skipping can speed up filtered queries when creating exact indexes isn’t possible.

AI-written summary · about

" chapters (8)

1. 0:00Introduction
2. 0:50What is data skipping
3. 4:44Columnar engine design
4. 15:25Data skipping in Tarantool
5. 24:23Existing data skipping approaches
6. 29:00Data skipping and logical predicates
7. 33:16Summary
8. 33:36Q&A

" related talks

• 47mStored Procedures in NoSQL DBMSs: The Tarantool ExampleDenis Linnik · HighLoad++ 2017
• 39mFrom PostgreSQL to Tarantool: Choosing When There Is No ChoiceEvgeniy Demkin, Lyubov Vaynshteyn · Joker 2022
• 49mPrimary Use Cases for In-Memory DBMS Illustrated by TarantoolDenis Anikin · РИТ++ 2016
• 23mSynchronizing Data from PgSQL to TarantoolVeniamin Gvozdikov · HighLoad++ 2017
• 29mDistributed Loose Index Scan (aka Skip/Jump/Hybrid Scan): Time Series in YugabyteDBFranck Pachot · Distributed SQL Summit Asia 2023
• 38mHow to Forget About Performance Problems? Tarantool as the DBMSAleksandr Kharchenko · Saint HighLoad++ 2022
• 46mTarantool: How to Save a Million Dollars on Your DBDenis Anikin · HighLoad++ 2015
• 49mIt’s Not Always About the Index: The Hidden Secrets of Query PerformanceAvishag Sahar · NDC Copenhagen 2025