Dive deep into the VertiPaq engine's architecture and functionality in this comprehensive 1-hour 10-minute conference talk from PASS Data Community Summit. Explore the columnar database used by SQL Server Analysis Services Tabular, Power BI, and Power Pivot, known for its impressive performance in speed and compression. Discover how VertiPaq stores information, gaining valuable insights into its inner workings and learning optimal data warehouse modeling techniques. Examine common and effective methods to enhance compression ratios and boost performance in Tabular data models. Cover topics such as row and column storage layouts, querying techniques, value encoding, hash encoding, Run Length Encoding (RLE), segmentation, processing phases, calculated columns and tables, query parallelization, hierarchies, memory usage, and materialization strategies.
Overview
Syllabus
Intro
Marco Russo
What is VertiPaq?
Tabular query architecture
Row storage layout
Column storage layout
Querying a columnar database
Column vs row storage
Value encoding
Hash encoding
Run Length Encoding (RLE)
Compression of one column
VertiPaq compression
Reducing column size
Segmentation
Processing phases
Special Case of 3rd Segment
Calculated columns and tables
Query parallelization
Hierarchies
Memory Usage During Process
Data memory usage
VertiPaq Analyzer
Query memory usage
Early materialization
Late materialization
Taught by
PASS Data Community Summit
