Oracle Exadata Performance on non-Oracle Cloud with Snowflake
Many Oracle customers do not wish, or cannot easily, migrate from Oracle but would like to be able to run Oracle Data Warehouse workloads faster and in a more cost-effective manner. With Snowflake and Gluent, it is possible to run Oracle Data Warehouses more economically and with similar or better performance than Oracle Exadata Machines. Gluent Data Platform allows you to migrate your Oracle workloads to Snowflake, running on all public cloud platforms (AWS, Azure and GCP), without creating any ETL or altering a single line of code in your existing applications.
Oracle Cloud Incompatibility
Oracle Exadata and Exadata Storage Servers are only available on premise or on Oracle Cloud. Until now, if you want to run Oracle applications with Oracle Exadata performance on AWS, Azure or GCP it cannot be done without a full migration to a different data warehouse, like Snowflake (good choice). Even if you are not an Exadata customer but are running Oracle RAC to get scale out compute, Oracle RAC is not supported either on AWS, Azure or GCP.
The Science of the Appliance
Oracle Exadata is an appliance optimised for running Oracle databases. One of the key ingredients of Oracle Exadata’s high performance is its use of Smart Scan. Smart Scan is a form of computation storage, supported by Exadata Storage Servers, where the storage has compute capability that can execute SQL projection (column) and predicate (filter) clauses in parallel and return only matching rows and columns rather than entire data blocks. This dramatically improves execution times and eliminates bottlenecks, because the network between storage and compute is only transmitting a set of matching data. The Oracle query optimizer is aware of the presence of Storage Servers that support Smart Scan and generates a different execution plan to leverage this push down capability.
Snowflake partner Gluent performs a similar process to Oracle Exadata by modifying the Oracle execution plan on the fly to push down query workloads to Snowflake returning an Oracle compatible resultset seamlessly. Gluent does not just push down filters and column projections it can also push down:
- Table Joins
- Prefiltered Table Joins
- Advanced Aggregations
This deep push down ability in combination with the high performance of Snowflake explains why Gluent and Snowflake can deliver greater performance. Leveraging Snowflake’s highly economical compute, storage and reduced administration further explains how this improved performance is achieved at lower cost.
What does this mean?
With Gluent and Snowflake you can get Oracle Exadata performance on AWS, Azure or GCP without having to migrate your Oracle SQL and PL/SQL. Your Oracle database front-end can be on premise or on cloud. Gluent implements hybrid queries where recent volatile data is queried from Oracle and older, higher volume and less volatile data is queried via Snowflake. Gluent is leveraging Snowflake as highly sophisticated computational storage and seamlessly integrating that into Oracle SQL execution plans.
Is this really a “no code” solution?
As close as you will get. The only command you need to run in Oracle to leverage Snowflake once Gluent is installed is:
SQL> alter session use <hybrid_schema>;All your Oracle queries then run unchanged.
The hybrid schema is built and maintained by Gluent to facilitate query rewrite and offload. Gluent also takes care of synchronising data from Oracle to Snowflake where you can then choose to purge the data that has been synchronised from Oracle.
What are the benefits?
The Gluent solution in combination with Snowflake allows Oracle customers to:
- Run Oracle workloads on AWS, Azure or GCP
- Run Hybrid Transactional and Analytics (HTAP) workloads leveraging the best of Oracle and Snowflake combined.
- Reduce Oracle footprint, enabling downsize from Exadata to RAC (on premise) or RAC to Single Instance (on cloud)
- Leverage Snowflake’s Data Sharing both for Publishing and Subscribing to datasets
- Optionally migrate from Oracle over time, typically starting with ELT
Our workload is special, will it work for me?
Gluent offers a free service, Gluent Advisor, which uses ASH/AWR data to determine how much data and processing can be moved to Snowflake from your unique environment. Find out more here — https://gluent.com/gdp/gluent-advisor
