Re: Comparing performance with Exadata

  • From: Lothar Flatz <l.flatz@xxxxxxxxxx>
  • To: learnerdatabase99@xxxxxxxxx, Oracle L <oracle-l@xxxxxxxxxxxxx>
  • Date: Thu, 22 Jun 2023 07:28:26 +0200

Hi,

I do not know Snowflake, but I did a number of POCs where Oracle/Exadata was compared against other hardware and databases. E.g Teradata, Netezza etc.
I can confirm that expertise is the most valuable asset in such POCs. If you just copy the data into the cloud and dump it on Snowflake, it is a sure way to see Snowflake failing.
Consider that your colleagues might favour that outcome.
Quite often companies will provide support for free if a deal is o stake.
You might ask Snowflake if they want to help on designing a suitable schema.

Thanks

Lothar

Am 21.06.2023 um 22:59 schrieb yudhi s:

Hello All,
We are mostly using Oracle Exadata(X9 and DB version 19C) and they are all on premises currently. These are currently supporting a variety of workloads  i.e OLTP + ETL type of workload , reporting application involving extensive use of plsql procedures and also UI search screens exposed directly to customers. So it's kind of supporting hybrid workloads now.

Few of the applications(OLAP Type) are moving to the cloud and management wanted to evaluate the Snowflake database for those. Wanted to compare its performance and suitability, pros/cons against Oracle exadata. But mainly wanted to compare the performance of these two.

We know Snowflake does not have indexes, user defined partitioning ,  but relies on micro partitions to filter out data , it doesn't have constraints etc. But they have separation of storage and compute i.e both can independently scale. It's columnar in nature etc. At the same time, we are also hearing of snowflake DB serving many critical OLAP /analytics type workloads.

Apology if this is silly one , but colleagues were thinking of just dumping/creating billions of rows in a table and then have read/write queries run on top of that table,  with similar structure in both the databases and comparing the response time. But i believe, as each database has their strengths/weaknesses so i am wondering if this simple way of testing would be really going to give us a fair result about the performance comparison. Please guide me here.

Want to understand from experts here , on how we should test and compare the performance of the databases in such cases to make any decision? Or anybody has experience using these databases as compared to Oracle Exadata?

Regards
Y

Other related posts: