Data Warehouse Showdown: Redshift vs BigQuery vs Snowflake

[@joachimzaspel8905]

Excellent! Mr Fraser gave a good presentation. Ease of use and in particular the amount of time necessary to do performance tuning for your data warehouse is a very important selection criteria. Ideally you have to do none and consistently get good enough results for different queries. Having to create and maintain indices yourself is a nightmare of the past. Modern analytical databases do all the self tuning behind the curtain. Really interesting are the benchmarks by Mr Litwintschik on BrytlytDB on a system with 20 P100 Nvidia GPU cards. One card costs 15.000$. So just the cards are 300.000$. I wonder how BrytlytDB would have performed on the TPC-DS dataset and queries - after all it is a fully relational database based on PostgreSql. GPUs for relational DBs seem to make a whole lot of sense - in particular for OLAP.

[@georgewfraser]

The Litwintschik benchmarks of BrytlytDB and other GPU data warehouses aren't terribly meaningful, because he doesn't count the time to load the data into GPU memory. Also, his queries are simple group-bys with no joins, which are not very representatitive of real-world queries, unless you are in 1985 and using an OLAP cube instead of a real data warehouse.

[@smallclips4164]

Please make a new video and this time compare azure as well. Microsoft had have been investing too much in BI and I am working on MSBI since 2004. Now i moved to aws but I still feel microsoft has done a pretty good job in cloud.

[@deepallep]

Great! maybe can add Alibaba Cloud MaxCompute next time, it is also a very competitive cloud data warehouse

[@NewSQL]

Derek, it looks like you represent Alibaba Cloud - please email us: data -at- DBMonth.com

[@Klayhamn]

Audio is horrible

[@NewSQL]

We have upgraded our audio-recording equipment since this was shot - this venue had 16-foot ceilings that echoed like crazy...

[@RaghuKalyanAnna]

@@NewSQL Benchmark related discussion could not follow properly. Can we see the slides and related SQL queries used in the benchmark tests?

[@rahulvermasir]

Try this kzbin.info/www/bejne/jqHEf2CGpriWsK8

[@georgefraser6310]

The latest version of our warehouse benchmark is at fivetran.com/blog/warehouse-benchmark

[@georgewfraser]

Latest version with Presto, Azure SQL DW, 1 TB scale is here: fivetran.com/blog/warehouse-benchmark