TPC-H Data Vault test on Exasol 7.0 beta

Contributor

Hi all,

wanted to share a small performance test, we made with the new Exasol 7.0 beta.

I was very curious how big the impact of the new Hash datatype would be.

 

Conditions:

- Exasol 6.2.7 community edition vs. Exasol 7.0 beta community edition

- both 4GB Ram

- TPC-H scalefactor 1

- persistent Raw Data Vault

- virtualized dimensional model

- query cache disabled

 

These are the results:

ADoerr_0-1594898820158.png

 

The performance improvement for the direct Raw DV queries is already really good,

but the improvement on the virtualized dimensional model is just suprising.

 

 

 

 

 

7 REPLIES 7

Contributor

 Thank you!  This is very encouraging.

Team Exasol
Team Exasol

Hi, we did indeed make similar experiments internally when developing Hashtype.

For that we created a data vault model using different datatypes for the keys and compared the TPC-H queries on view layer on top of these different models.

sf25.png

sf100.png

Xpert

Did you use the new optimizer or the "old"?

Contributor

 

I did use the "old" one.
My main intention was to test the influence of new Hash datatype.

Xpert

Hi ADoerr,

this looks really cool ! Did you also test the rest of Q1 - Q22 from the test set and would be able to share the results ? The more data the better 😉

Cheers
Malte

Contributor

 

No, I just used the mentioned 2 queries at first as this was for a small webinar demo.

Maybe I will get bored at the weekend and test a little more.

Team Exasol
Team Exasol

Maybe @exa-StefanM can share something 😉