Tableau

Row-store have all associated data in a tuple, making it easy to add/modify records
Column-stores have all data points as separate entries, storing them individually meaning we won't make unecessary reads
provides heavy optimization for complex queries on wide databases
- less reads since we might only read a few columns instead of hundreds
- column-stores can optimize for compression (more consistent formatting so less entropy)
- sorted columns compress better, range queries are faster (since the data can be stored sequentially)
- performance benefits from clustered indexing

Tableau Data Extracts are a compressed set of files that represent a data table from our data source, for the purpose of visualization in Tableau.

TDE's are heavily optimized to work with large datasets using low RAM and reduced Disk
TDE's have a file for each column (file of data elements) + metadata file(s)
TDE is a single file that contains many individual memory-mapped files, composed of column and metadata files
compression is not general file compression, it's run length encoding and dictionary compression so no need for decompression when loading into memory

Compressed snapshot of data stored on disk - saved subsets of data sources for performance optimization

can add functionality to data filtering
extracts can perform full refreshes or incremental refreshes that add rows that are new since the previous refresh
when a data extract is created, the structure for the extract is first defined, with a searate file for each column in the underling source
TDE compression is separate from file compression, no decompression to load into memory
- uses dictionar compression and run length encoding

network speeds between machines are a bottleneck
server hosted extracts will be faster
Individual row pull speed may still be an issue, depending on the query setup (Subqueries in the SELECT items, for example). Try landing the data to a table temporarily (or permanently, reloaded with an sproc) to see the speed difference.
Bringing down the number of columns, especially text columns will definitely speed up the query significantly.