How SlicingDice performs on the TPC-DS benchmark

We used all scripts and the TPC-DS datasets prepared by the Fivetran's team and available on this Github repository. This TPC-DS dataset is composed of 24 tables in a star schema; the tables represent web, catalog, and store sales of an imaginary retailer. The largest fact table has 400 million rows at 100GB scale and 4 billion rows at 1TB scale.

As we don't disput the results presented by Fivetran for Redshift, BigQuery and Snowflake, we simply ran all the 99 TPC-DS queries on SlicingDice and added the results to this comparison table below. Query time presented with geometric mean.

Query Speed

Query Speed

100 GB Dataset (400 million rows)

1 TB Dataset (4 billion rows)


Infrastructure

100 GB Dataset (400 million rows)

1 TB Dataset (4 billion rows)


Monthly Cost

100 GB Dataset
(400 million rows)

1 TB Dataset
(4 billion rows)

...

 

1.15 Seconds

2.53 Seconds


 

Infinite

Infinite


 

$25.00
($0.03/hour)

$250.00
($0.34/hour)

...

 

19.45 Seconds

20.50 Seconds


 

8 × dc2.large

4 × dc2.8xlarge


 

$1,440.00
($2.00/hour)

$13,824.00
($19.20/hour)

...

 

8.10 Seconds

14.32 Seconds


 

Unknown

Unknown


 


Unpredictable


Unpredictable

...

 

6.12 Seconds

10.74 Seconds


 

X-Small

Large


 

$1,440.00
($2.00/hour)

$11,520.00
($16.00/hour)



Do you want to reproduce this performance test using your own SlicingDice account? Feel free to take this challenge and check by yourself how SlicingDice can surprise you. Talk to our team to help set you up.

Ready to get started?

Get in touch, or create an account.